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 3.13 /proc/<pid>/fd - List of symlinks to open files 51 52 4 Configuring procfs 53 4.1 Mount options 54 55 5 Filesystem behavior 56 57Preface 58======= 59 600.1 Introduction/Credits 61------------------------ 62 63This documentation is part of a soon (or so we hope) to be released book on 64the SuSE Linux distribution. As there is no complete documentation for the 65/proc file system and we've used many freely available sources to write these 66chapters, it seems only fair to give the work back to the Linux community. 67This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm 68afraid it's still far from complete, but we hope it will be useful. As far as 69we know, it is the first 'all-in-one' document about the /proc file system. It 70is focused on the Intel x86 hardware, so if you are looking for PPC, ARM, 71SPARC, AXP, etc., features, you probably won't find what you are looking for. 72It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But 73additions and patches are welcome and will be added to this document if you 74mail them to Bodo. 75 76We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of 77other people for help compiling this documentation. We'd also like to extend a 78special thank you to Andi Kleen for documentation, which we relied on heavily 79to create this document, as well as the additional information he provided. 80Thanks to everybody else who contributed source or docs to the Linux kernel 81and helped create a great piece of software... :) 82 83If you have any comments, corrections or additions, please don't hesitate to 84contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this 85document. 86 87The latest version of this document is available online at 88http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html 89 90If the above direction does not works for you, you could try the kernel 91mailing list at linux-kernel@vger.kernel.org and/or try to reach me at 92comandante@zaralinux.com. 93 940.2 Legal Stuff 95--------------- 96 97We don't guarantee the correctness of this document, and if you come to us 98complaining about how you screwed up your system because of incorrect 99documentation, we won't feel responsible... 100 101Chapter 1: Collecting System Information 102======================================== 103 104In This Chapter 105--------------- 106* Investigating the properties of the pseudo file system /proc and its 107 ability to provide information on the running Linux system 108* Examining /proc's structure 109* Uncovering various information about the kernel and the processes running 110 on the system 111 112------------------------------------------------------------------------------ 113 114The proc file system acts as an interface to internal data structures in the 115kernel. It can be used to obtain information about the system and to change 116certain kernel parameters at runtime (sysctl). 117 118First, we'll take a look at the read-only parts of /proc. In Chapter 2, we 119show you how you can use /proc/sys to change settings. 120 1211.1 Process-Specific Subdirectories 122----------------------------------- 123 124The directory /proc contains (among other things) one subdirectory for each 125process running on the system, which is named after the process ID (PID). 126 127The link 'self' points to the process reading the file system. Each process 128subdirectory has the entries listed in Table 1-1. 129 130Note that an open file descriptor to /proc/<pid> or to any of its 131contained files or subdirectories does not prevent <pid> being reused 132for some other process in the event that <pid> exits. Operations on 133open /proc/<pid> file descriptors corresponding to dead processes 134never act on any new process that the kernel may, through chance, have 135also assigned the process ID <pid>. Instead, operations on these FDs 136usually fail with ESRCH. 137 138.. table:: Table 1-1: Process specific entries in /proc 139 140 ============= =============================================================== 141 File Content 142 ============= =============================================================== 143 clear_refs Clears page referenced bits shown in smaps output 144 cmdline Command line arguments 145 cpu Current and last cpu in which it was executed (2.4)(smp) 146 cwd Link to the current working directory 147 environ Values of environment variables 148 exe Link to the executable of this process 149 fd Directory, which contains all file descriptors 150 maps Memory maps to executables and library files (2.4) 151 mem Memory held by this process 152 root Link to the root directory of this process 153 stat Process status 154 statm Process memory status information 155 status Process status in human readable form 156 wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function 157 symbol the task is blocked in - or "0" if not blocked. 158 pagemap Page table 159 stack Report full stack trace, enable via CONFIG_STACKTRACE 160 smaps An extension based on maps, showing the memory consumption of 161 each mapping and flags associated with it 162 smaps_rollup Accumulated smaps stats for all mappings of the process. This 163 can be derived from smaps, but is faster and more convenient 164 numa_maps An extension based on maps, showing the memory locality and 165 binding policy as well as mem usage (in pages) of each mapping. 166 ============= =============================================================== 167 168For example, to get the status information of a process, all you have to do is 169read the file /proc/PID/status:: 170 171 >cat /proc/self/status 172 Name: cat 173 State: R (running) 174 Tgid: 5452 175 Pid: 5452 176 PPid: 743 177 TracerPid: 0 (2.4) 178 Uid: 501 501 501 501 179 Gid: 100 100 100 100 180 FDSize: 256 181 Groups: 100 14 16 182 VmPeak: 5004 kB 183 VmSize: 5004 kB 184 VmLck: 0 kB 185 VmHWM: 476 kB 186 VmRSS: 476 kB 187 RssAnon: 352 kB 188 RssFile: 120 kB 189 RssShmem: 4 kB 190 VmData: 156 kB 191 VmStk: 88 kB 192 VmExe: 68 kB 193 VmLib: 1412 kB 194 VmPTE: 20 kb 195 VmSwap: 0 kB 196 HugetlbPages: 0 kB 197 CoreDumping: 0 198 THP_enabled: 1 199 Threads: 1 200 SigQ: 0/28578 201 SigPnd: 0000000000000000 202 ShdPnd: 0000000000000000 203 SigBlk: 0000000000000000 204 SigIgn: 0000000000000000 205 SigCgt: 0000000000000000 206 CapInh: 00000000fffffeff 207 CapPrm: 0000000000000000 208 CapEff: 0000000000000000 209 CapBnd: ffffffffffffffff 210 CapAmb: 0000000000000000 211 NoNewPrivs: 0 212 Seccomp: 0 213 Speculation_Store_Bypass: thread vulnerable 214 SpeculationIndirectBranch: conditional enabled 215 voluntary_ctxt_switches: 0 216 nonvoluntary_ctxt_switches: 1 217 218This shows you nearly the same information you would get if you viewed it with 219the ps command. In fact, ps uses the proc file system to obtain its 220information. But you get a more detailed view of the process by reading the 221file /proc/PID/status. It fields are described in table 1-2. 222 223The statm file contains more detailed information about the process 224memory usage. Its seven fields are explained in Table 1-3. The stat file 225contains detailed information about the process itself. Its fields are 226explained in Table 1-4. 227 228(for SMP CONFIG users) 229 230For making accounting scalable, RSS related information are handled in an 231asynchronous manner and the value may not be very precise. To see a precise 232snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table. 233It's slow but very precise. 234 235.. table:: Table 1-2: Contents of the status files (as of 4.19) 236 237 ========================== =================================================== 238 Field Content 239 ========================== =================================================== 240 Name filename of the executable 241 Umask file mode creation mask 242 State state (R is running, S is sleeping, D is sleeping 243 in an uninterruptible wait, Z is zombie, 244 T is traced or stopped) 245 Tgid thread group ID 246 Ngid NUMA group ID (0 if none) 247 Pid process id 248 PPid process id of the parent process 249 TracerPid PID of process tracing this process (0 if not, or 250 the tracer is outside of the current pid namespace) 251 Uid Real, effective, saved set, and file system UIDs 252 Gid Real, effective, saved set, and file system GIDs 253 FDSize number of file descriptor slots currently allocated 254 Groups supplementary group list 255 NStgid descendant namespace thread group ID hierarchy 256 NSpid descendant namespace process ID hierarchy 257 NSpgid descendant namespace process group ID hierarchy 258 NSsid descendant namespace session ID hierarchy 259 VmPeak peak virtual memory size 260 VmSize total program size 261 VmLck locked memory size 262 VmPin pinned memory size 263 VmHWM peak resident set size ("high water mark") 264 VmRSS size of memory portions. It contains the three 265 following parts 266 (VmRSS = RssAnon + RssFile + RssShmem) 267 RssAnon size of resident anonymous memory 268 RssFile size of resident file mappings 269 RssShmem size of resident shmem memory (includes SysV shm, 270 mapping of tmpfs and shared anonymous mappings) 271 VmData size of private data segments 272 VmStk size of stack segments 273 VmExe size of text segment 274 VmLib size of shared library code 275 VmPTE size of page table entries 276 VmSwap amount of swap used by anonymous private data 277 (shmem swap usage is not included) 278 HugetlbPages size of hugetlb memory portions 279 CoreDumping process's memory is currently being dumped 280 (killing the process may lead to a corrupted core) 281 THP_enabled process is allowed to use THP (returns 0 when 282 PR_SET_THP_DISABLE is set on the process 283 Threads number of threads 284 SigQ number of signals queued/max. number for queue 285 SigPnd bitmap of pending signals for the thread 286 ShdPnd bitmap of shared pending signals for the process 287 SigBlk bitmap of blocked signals 288 SigIgn bitmap of ignored signals 289 SigCgt bitmap of caught signals 290 CapInh bitmap of inheritable capabilities 291 CapPrm bitmap of permitted capabilities 292 CapEff bitmap of effective capabilities 293 CapBnd bitmap of capabilities bounding set 294 CapAmb bitmap of ambient capabilities 295 NoNewPrivs no_new_privs, like prctl(PR_GET_NO_NEW_PRIV, ...) 296 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...) 297 Speculation_Store_Bypass speculative store bypass mitigation status 298 SpeculationIndirectBranch indirect branch speculation mode 299 Cpus_allowed mask of CPUs on which this process may run 300 Cpus_allowed_list Same as previous, but in "list format" 301 Mems_allowed mask of memory nodes allowed to this process 302 Mems_allowed_list Same as previous, but in "list format" 303 voluntary_ctxt_switches number of voluntary context switches 304 nonvoluntary_ctxt_switches number of non voluntary context switches 305 ========================== =================================================== 306 307 308.. table:: Table 1-3: Contents of the statm files (as of 2.6.8-rc3) 309 310 ======== =============================== ============================== 311 Field Content 312 ======== =============================== ============================== 313 size total program size (pages) (same as VmSize in status) 314 resident size of memory portions (pages) (same as VmRSS in status) 315 shared number of pages that are shared (i.e. backed by a file, same 316 as RssFile+RssShmem in status) 317 trs number of pages that are 'code' (not including libs; broken, 318 includes data segment) 319 lrs number of pages of library (always 0 on 2.6) 320 drs number of pages of data/stack (including libs; broken, 321 includes library text) 322 dt number of dirty pages (always 0 on 2.6) 323 ======== =============================== ============================== 324 325 326.. table:: Table 1-4: Contents of the stat files (as of 2.6.30-rc7) 327 328 ============= =============================================================== 329 Field Content 330 ============= =============================================================== 331 pid process id 332 tcomm filename of the executable 333 state state (R is running, S is sleeping, D is sleeping in an 334 uninterruptible wait, Z is zombie, T is traced or stopped) 335 ppid process id of the parent process 336 pgrp pgrp of the process 337 sid session id 338 tty_nr tty the process uses 339 tty_pgrp pgrp of the tty 340 flags task flags 341 min_flt number of minor faults 342 cmin_flt number of minor faults with child's 343 maj_flt number of major faults 344 cmaj_flt number of major faults with child's 345 utime user mode jiffies 346 stime kernel mode jiffies 347 cutime user mode jiffies with child's 348 cstime kernel mode jiffies with child's 349 priority priority level 350 nice nice level 351 num_threads number of threads 352 it_real_value (obsolete, always 0) 353 start_time time the process started after system boot 354 vsize virtual memory size 355 rss resident set memory size 356 rsslim current limit in bytes on the rss 357 start_code address above which program text can run 358 end_code address below which program text can run 359 start_stack address of the start of the main process stack 360 esp current value of ESP 361 eip current value of EIP 362 pending bitmap of pending signals 363 blocked bitmap of blocked signals 364 sigign bitmap of ignored signals 365 sigcatch bitmap of caught signals 366 0 (place holder, used to be the wchan address, 367 use /proc/PID/wchan instead) 368 0 (place holder) 369 0 (place holder) 370 exit_signal signal to send to parent thread on exit 371 task_cpu which CPU the task is scheduled on 372 rt_priority realtime priority 373 policy scheduling policy (man sched_setscheduler) 374 blkio_ticks time spent waiting for block IO 375 gtime guest time of the task in jiffies 376 cgtime guest time of the task children in jiffies 377 start_data address above which program data+bss is placed 378 end_data address below which program data+bss is placed 379 start_brk address above which program heap can be expanded with brk() 380 arg_start address above which program command line is placed 381 arg_end address below which program command line is placed 382 env_start address above which program environment is placed 383 env_end address below which program environment is placed 384 exit_code the thread's exit_code in the form reported by the waitpid 385 system call 386 ============= =============================================================== 387 388The /proc/PID/maps file contains the currently mapped memory regions and 389their access permissions. 390 391The format is:: 392 393 address perms offset dev inode pathname 394 395 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test 396 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test 397 0804a000-0806b000 rw-p 00000000 00:00 0 [heap] 398 a7cb1000-a7cb2000 ---p 00000000 00:00 0 399 a7cb2000-a7eb2000 rw-p 00000000 00:00 0 400 a7eb2000-a7eb3000 ---p 00000000 00:00 0 401 a7eb3000-a7ed5000 rw-p 00000000 00:00 0 402 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6 403 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6 404 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6 405 a800b000-a800e000 rw-p 00000000 00:00 0 406 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0 407 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0 408 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0 409 a8024000-a8027000 rw-p 00000000 00:00 0 410 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2 411 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2 412 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2 413 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack] 414 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso] 415 416where "address" is the address space in the process that it occupies, "perms" 417is a set of permissions:: 418 419 r = read 420 w = write 421 x = execute 422 s = shared 423 p = private (copy on write) 424 425"offset" is the offset into the mapping, "dev" is the device (major:minor), and 426"inode" is the inode on that device. 0 indicates that no inode is associated 427with the memory region, as the case would be with BSS (uninitialized data). 428The "pathname" shows the name associated file for this mapping. If the mapping 429is not associated with a file: 430 431 =================== =========================================== 432 [heap] the heap of the program 433 [stack] the stack of the main process 434 [vdso] the "virtual dynamic shared object", 435 the kernel system call handler 436 [anon:<name>] a private anonymous mapping that has been 437 named by userspace 438 [anon_shmem:<name>] an anonymous shared memory mapping that has 439 been named by userspace 440 =================== =========================================== 441 442 or if empty, the mapping is anonymous. 443 444The /proc/PID/smaps is an extension based on maps, showing the memory 445consumption for each of the process's mappings. For each mapping (aka Virtual 446Memory Area, or VMA) there is a series of lines such as the following:: 447 448 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash 449 450 Size: 1084 kB 451 KernelPageSize: 4 kB 452 MMUPageSize: 4 kB 453 Rss: 892 kB 454 Pss: 374 kB 455 Pss_Dirty: 0 kB 456 Shared_Clean: 892 kB 457 Shared_Dirty: 0 kB 458 Private_Clean: 0 kB 459 Private_Dirty: 0 kB 460 Referenced: 892 kB 461 Anonymous: 0 kB 462 LazyFree: 0 kB 463 AnonHugePages: 0 kB 464 ShmemPmdMapped: 0 kB 465 Shared_Hugetlb: 0 kB 466 Private_Hugetlb: 0 kB 467 Swap: 0 kB 468 SwapPss: 0 kB 469 KernelPageSize: 4 kB 470 MMUPageSize: 4 kB 471 Locked: 0 kB 472 THPeligible: 0 473 VmFlags: rd ex mr mw me dw 474 475The first of these lines shows the same information as is displayed for the 476mapping in /proc/PID/maps. Following lines show the size of the mapping 477(size); the size of each page allocated when backing a VMA (KernelPageSize), 478which is usually the same as the size in the page table entries; the page size 479used by the MMU when backing a VMA (in most cases, the same as KernelPageSize); 480the amount of the mapping that is currently resident in RAM (RSS); the 481process' proportional share of this mapping (PSS); and the number of clean and 482dirty shared and private pages in the mapping. 483 484The "proportional set size" (PSS) of a process is the count of pages it has 485in memory, where each page is divided by the number of processes sharing it. 486So if a process has 1000 pages all to itself, and 1000 shared with one other 487process, its PSS will be 1500. "Pss_Dirty" is the portion of PSS which 488consists of dirty pages. ("Pss_Clean" is not included, but it can be 489calculated by subtracting "Pss_Dirty" from "Pss".) 490 491Note that even a page which is part of a MAP_SHARED mapping, but has only 492a single pte mapped, i.e. is currently used by only one process, is accounted 493as private and not as shared. 494 495"Referenced" indicates the amount of memory currently marked as referenced or 496accessed. 497 498"Anonymous" shows the amount of memory that does not belong to any file. Even 499a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE 500and a page is modified, the file page is replaced by a private anonymous copy. 501 502"LazyFree" shows the amount of memory which is marked by madvise(MADV_FREE). 503The memory isn't freed immediately with madvise(). It's freed in memory 504pressure if the memory is clean. Please note that the printed value might 505be lower than the real value due to optimizations used in the current 506implementation. If this is not desirable please file a bug report. 507 508"AnonHugePages" shows the ammount of memory backed by transparent hugepage. 509 510"ShmemPmdMapped" shows the ammount of shared (shmem/tmpfs) memory backed by 511huge pages. 512 513"Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by 514hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical 515reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field. 516 517"Swap" shows how much would-be-anonymous memory is also used, but out on swap. 518 519For shmem mappings, "Swap" includes also the size of the mapped (and not 520replaced by copy-on-write) part of the underlying shmem object out on swap. 521"SwapPss" shows proportional swap share of this mapping. Unlike "Swap", this 522does not take into account swapped out page of underlying shmem objects. 523"Locked" indicates whether the mapping is locked in memory or not. 524 525"THPeligible" indicates whether the mapping is eligible for allocating THP 526pages as well as the THP is PMD mappable or not - 1 if true, 0 otherwise. 527It just shows the current status. 528 529"VmFlags" field deserves a separate description. This member represents the 530kernel flags associated with the particular virtual memory area in two letter 531encoded manner. The codes are the following: 532 533 == ======================================= 534 rd readable 535 wr writeable 536 ex executable 537 sh shared 538 mr may read 539 mw may write 540 me may execute 541 ms may share 542 gd stack segment growns down 543 pf pure PFN range 544 dw disabled write to the mapped file 545 lo pages are locked in memory 546 io memory mapped I/O area 547 sr sequential read advise provided 548 rr random read advise provided 549 dc do not copy area on fork 550 de do not expand area on remapping 551 ac area is accountable 552 nr swap space is not reserved for the area 553 ht area uses huge tlb pages 554 sf synchronous page fault 555 ar architecture specific flag 556 wf wipe on fork 557 dd do not include area into core dump 558 sd soft dirty flag 559 mm mixed map area 560 hg huge page advise flag 561 nh no huge page advise flag 562 mg mergable advise flag 563 bt arm64 BTI guarded page 564 mt arm64 MTE allocation tags are enabled 565 um userfaultfd missing tracking 566 uw userfaultfd wr-protect tracking 567 == ======================================= 568 569Note that there is no guarantee that every flag and associated mnemonic will 570be present in all further kernel releases. Things get changed, the flags may 571be vanished or the reverse -- new added. Interpretation of their meaning 572might change in future as well. So each consumer of these flags has to 573follow each specific kernel version for the exact semantic. 574 575This file is only present if the CONFIG_MMU kernel configuration option is 576enabled. 577 578Note: reading /proc/PID/maps or /proc/PID/smaps is inherently racy (consistent 579output can be achieved only in the single read call). 580 581This typically manifests when doing partial reads of these files while the 582memory map is being modified. Despite the races, we do provide the following 583guarantees: 584 5851) The mapped addresses never go backwards, which implies no two 586 regions will ever overlap. 5872) If there is something at a given vaddr during the entirety of the 588 life of the smaps/maps walk, there will be some output for it. 589 590The /proc/PID/smaps_rollup file includes the same fields as /proc/PID/smaps, 591but their values are the sums of the corresponding values for all mappings of 592the process. Additionally, it contains these fields: 593 594- Pss_Anon 595- Pss_File 596- Pss_Shmem 597 598They represent the proportional shares of anonymous, file, and shmem pages, as 599described for smaps above. These fields are omitted in smaps since each 600mapping identifies the type (anon, file, or shmem) of all pages it contains. 601Thus all information in smaps_rollup can be derived from smaps, but at a 602significantly higher cost. 603 604The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG 605bits on both physical and virtual pages associated with a process, and the 606soft-dirty bit on pte (see Documentation/admin-guide/mm/soft-dirty.rst 607for details). 608To clear the bits for all the pages associated with the process:: 609 610 > echo 1 > /proc/PID/clear_refs 611 612To clear the bits for the anonymous pages associated with the process:: 613 614 > echo 2 > /proc/PID/clear_refs 615 616To clear the bits for the file mapped pages associated with the process:: 617 618 > echo 3 > /proc/PID/clear_refs 619 620To clear the soft-dirty bit:: 621 622 > echo 4 > /proc/PID/clear_refs 623 624To reset the peak resident set size ("high water mark") to the process's 625current value:: 626 627 > echo 5 > /proc/PID/clear_refs 628 629Any other value written to /proc/PID/clear_refs will have no effect. 630 631The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags 632using /proc/kpageflags and number of times a page is mapped using 633/proc/kpagecount. For detailed explanation, see 634Documentation/admin-guide/mm/pagemap.rst. 635 636The /proc/pid/numa_maps is an extension based on maps, showing the memory 637locality and binding policy, as well as the memory usage (in pages) of 638each mapping. The output follows a general format where mapping details get 639summarized separated by blank spaces, one mapping per each file line:: 640 641 address policy mapping details 642 643 00400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4 644 00600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4 645 3206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4 646 320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 647 3206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 648 3206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4 649 3206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4 650 320698b000 default file=/lib64/libc-2.12.so 651 3206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4 652 3206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 653 3206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4 654 7f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4 655 7f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4 656 7f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048 657 7fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4 658 7fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4 659 660Where: 661 662"address" is the starting address for the mapping; 663 664"policy" reports the NUMA memory policy set for the mapping (see Documentation/admin-guide/mm/numa_memory_policy.rst); 665 666"mapping details" summarizes mapping data such as mapping type, page usage counters, 667node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page 668size, in KB, that is backing the mapping up. 669 6701.2 Kernel data 671--------------- 672 673Similar to the process entries, the kernel data files give information about 674the running kernel. The files used to obtain this information are contained in 675/proc and are listed in Table 1-5. Not all of these will be present in your 676system. It depends on the kernel configuration and the loaded modules, which 677files are there, and which are missing. 678 679.. table:: Table 1-5: Kernel info in /proc 680 681 ============ =============================================================== 682 File Content 683 ============ =============================================================== 684 apm Advanced power management info 685 buddyinfo Kernel memory allocator information (see text) (2.5) 686 bus Directory containing bus specific information 687 cmdline Kernel command line 688 cpuinfo Info about the CPU 689 devices Available devices (block and character) 690 dma Used DMS channels 691 filesystems Supported filesystems 692 driver Various drivers grouped here, currently rtc (2.4) 693 execdomains Execdomains, related to security (2.4) 694 fb Frame Buffer devices (2.4) 695 fs File system parameters, currently nfs/exports (2.4) 696 ide Directory containing info about the IDE subsystem 697 interrupts Interrupt usage 698 iomem Memory map (2.4) 699 ioports I/O port usage 700 irq Masks for irq to cpu affinity (2.4)(smp?) 701 isapnp ISA PnP (Plug&Play) Info (2.4) 702 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4)) 703 kmsg Kernel messages 704 ksyms Kernel symbol table 705 loadavg Load average of last 1, 5 & 15 minutes; 706 number of processes currently runnable (running or on ready queue); 707 total number of processes in system; 708 last pid created. 709 All fields are separated by one space except "number of 710 processes currently runnable" and "total number of processes 711 in system", which are separated by a slash ('/'). Example: 712 0.61 0.61 0.55 3/828 22084 713 locks Kernel locks 714 meminfo Memory info 715 misc Miscellaneous 716 modules List of loaded modules 717 mounts Mounted filesystems 718 net Networking info (see text) 719 pagetypeinfo Additional page allocator information (see text) (2.5) 720 partitions Table of partitions known to the system 721 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/, 722 decoupled by lspci (2.4) 723 rtc Real time clock 724 scsi SCSI info (see text) 725 slabinfo Slab pool info 726 softirqs softirq usage 727 stat Overall statistics 728 swaps Swap space utilization 729 sys See chapter 2 730 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4) 731 tty Info of tty drivers 732 uptime Wall clock since boot, combined idle time of all cpus 733 version Kernel version 734 video bttv info of video resources (2.4) 735 vmallocinfo Show vmalloced areas 736 ============ =============================================================== 737 738You can, for example, check which interrupts are currently in use and what 739they are used for by looking in the file /proc/interrupts:: 740 741 > cat /proc/interrupts 742 CPU0 743 0: 8728810 XT-PIC timer 744 1: 895 XT-PIC keyboard 745 2: 0 XT-PIC cascade 746 3: 531695 XT-PIC aha152x 747 4: 2014133 XT-PIC serial 748 5: 44401 XT-PIC pcnet_cs 749 8: 2 XT-PIC rtc 750 11: 8 XT-PIC i82365 751 12: 182918 XT-PIC PS/2 Mouse 752 13: 1 XT-PIC fpu 753 14: 1232265 XT-PIC ide0 754 15: 7 XT-PIC ide1 755 NMI: 0 756 757In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the 758output of a SMP machine):: 759 760 > cat /proc/interrupts 761 762 CPU0 CPU1 763 0: 1243498 1214548 IO-APIC-edge timer 764 1: 8949 8958 IO-APIC-edge keyboard 765 2: 0 0 XT-PIC cascade 766 5: 11286 10161 IO-APIC-edge soundblaster 767 8: 1 0 IO-APIC-edge rtc 768 9: 27422 27407 IO-APIC-edge 3c503 769 12: 113645 113873 IO-APIC-edge PS/2 Mouse 770 13: 0 0 XT-PIC fpu 771 14: 22491 24012 IO-APIC-edge ide0 772 15: 2183 2415 IO-APIC-edge ide1 773 17: 30564 30414 IO-APIC-level eth0 774 18: 177 164 IO-APIC-level bttv 775 NMI: 2457961 2457959 776 LOC: 2457882 2457881 777 ERR: 2155 778 779NMI is incremented in this case because every timer interrupt generates a NMI 780(Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups. 781 782LOC is the local interrupt counter of the internal APIC of every CPU. 783 784ERR is incremented in the case of errors in the IO-APIC bus (the bus that 785connects the CPUs in a SMP system. This means that an error has been detected, 786the IO-APIC automatically retry the transmission, so it should not be a big 787problem, but you should read the SMP-FAQ. 788 789In 2.6.2* /proc/interrupts was expanded again. This time the goal was for 790/proc/interrupts to display every IRQ vector in use by the system, not 791just those considered 'most important'. The new vectors are: 792 793THR 794 interrupt raised when a machine check threshold counter 795 (typically counting ECC corrected errors of memory or cache) exceeds 796 a configurable threshold. Only available on some systems. 797 798TRM 799 a thermal event interrupt occurs when a temperature threshold 800 has been exceeded for the CPU. This interrupt may also be generated 801 when the temperature drops back to normal. 802 803SPU 804 a spurious interrupt is some interrupt that was raised then lowered 805 by some IO device before it could be fully processed by the APIC. Hence 806 the APIC sees the interrupt but does not know what device it came from. 807 For this case the APIC will generate the interrupt with a IRQ vector 808 of 0xff. This might also be generated by chipset bugs. 809 810RES, CAL, TLB 811 rescheduling, call and TLB flush interrupts are 812 sent from one CPU to another per the needs of the OS. Typically, 813 their statistics are used by kernel developers and interested users to 814 determine the occurrence of interrupts of the given type. 815 816The above IRQ vectors are displayed only when relevant. For example, 817the threshold vector does not exist on x86_64 platforms. Others are 818suppressed when the system is a uniprocessor. As of this writing, only 819i386 and x86_64 platforms support the new IRQ vector displays. 820 821Of some interest is the introduction of the /proc/irq directory to 2.4. 822It could be used to set IRQ to CPU affinity. This means that you can "hook" an 823IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the 824irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and 825prof_cpu_mask. 826 827For example:: 828 829 > ls /proc/irq/ 830 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask 831 1 11 13 15 17 19 3 5 7 9 default_smp_affinity 832 > ls /proc/irq/0/ 833 smp_affinity 834 835smp_affinity is a bitmask, in which you can specify which CPUs can handle the 836IRQ. You can set it by doing:: 837 838 > echo 1 > /proc/irq/10/smp_affinity 839 840This means that only the first CPU will handle the IRQ, but you can also echo 8415 which means that only the first and third CPU can handle the IRQ. 842 843The contents of each smp_affinity file is the same by default:: 844 845 > cat /proc/irq/0/smp_affinity 846 ffffffff 847 848There is an alternate interface, smp_affinity_list which allows specifying 849a CPU range instead of a bitmask:: 850 851 > cat /proc/irq/0/smp_affinity_list 852 1024-1031 853 854The default_smp_affinity mask applies to all non-active IRQs, which are the 855IRQs which have not yet been allocated/activated, and hence which lack a 856/proc/irq/[0-9]* directory. 857 858The node file on an SMP system shows the node to which the device using the IRQ 859reports itself as being attached. This hardware locality information does not 860include information about any possible driver locality preference. 861 862prof_cpu_mask specifies which CPUs are to be profiled by the system wide 863profiler. Default value is ffffffff (all CPUs if there are only 32 of them). 864 865The way IRQs are routed is handled by the IO-APIC, and it's Round Robin 866between all the CPUs which are allowed to handle it. As usual the kernel has 867more info than you and does a better job than you, so the defaults are the 868best choice for almost everyone. [Note this applies only to those IO-APIC's 869that support "Round Robin" interrupt distribution.] 870 871There are three more important subdirectories in /proc: net, scsi, and sys. 872The general rule is that the contents, or even the existence of these 873directories, depend on your kernel configuration. If SCSI is not enabled, the 874directory scsi may not exist. The same is true with the net, which is there 875only when networking support is present in the running kernel. 876 877The slabinfo file gives information about memory usage at the slab level. 878Linux uses slab pools for memory management above page level in version 2.2. 879Commonly used objects have their own slab pool (such as network buffers, 880directory cache, and so on). 881 882:: 883 884 > cat /proc/buddyinfo 885 886 Node 0, zone DMA 0 4 5 4 4 3 ... 887 Node 0, zone Normal 1 0 0 1 101 8 ... 888 Node 0, zone HighMem 2 0 0 1 1 0 ... 889 890External fragmentation is a problem under some workloads, and buddyinfo is a 891useful tool for helping diagnose these problems. Buddyinfo will give you a 892clue as to how big an area you can safely allocate, or why a previous 893allocation failed. 894 895Each column represents the number of pages of a certain order which are 896available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in 897ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE 898available in ZONE_NORMAL, etc... 899 900More information relevant to external fragmentation can be found in 901pagetypeinfo:: 902 903 > cat /proc/pagetypeinfo 904 Page block order: 9 905 Pages per block: 512 906 907 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10 908 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0 909 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0 910 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2 911 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0 912 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0 913 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9 914 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0 915 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452 916 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0 917 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0 918 919 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate 920 Node 0, zone DMA 2 0 5 1 0 921 Node 0, zone DMA32 41 6 967 2 0 922 923Fragmentation avoidance in the kernel works by grouping pages of different 924migrate types into the same contiguous regions of memory called page blocks. 925A page block is typically the size of the default hugepage size, e.g. 2MB on 926X86-64. By keeping pages grouped based on their ability to move, the kernel 927can reclaim pages within a page block to satisfy a high-order allocation. 928 929The pagetypinfo begins with information on the size of a page block. It 930then gives the same type of information as buddyinfo except broken down 931by migrate-type and finishes with details on how many page blocks of each 932type exist. 933 934If min_free_kbytes has been tuned correctly (recommendations made by hugeadm 935from libhugetlbfs https://github.com/libhugetlbfs/libhugetlbfs/), one can 936make an estimate of the likely number of huge pages that can be allocated 937at a given point in time. All the "Movable" blocks should be allocatable 938unless memory has been mlock()'d. Some of the Reclaimable blocks should 939also be allocatable although a lot of filesystem metadata may have to be 940reclaimed to achieve this. 941 942 943meminfo 944~~~~~~~ 945 946Provides information about distribution and utilization of memory. This 947varies by architecture and compile options. Some of the counters reported 948here overlap. The memory reported by the non overlapping counters may not 949add up to the overall memory usage and the difference for some workloads 950can be substantial. In many cases there are other means to find out 951additional memory using subsystem specific interfaces, for instance 952/proc/net/sockstat for TCP memory allocations. 953 954Example output. You may not have all of these fields. 955 956:: 957 958 > cat /proc/meminfo 959 960 MemTotal: 32858820 kB 961 MemFree: 21001236 kB 962 MemAvailable: 27214312 kB 963 Buffers: 581092 kB 964 Cached: 5587612 kB 965 SwapCached: 0 kB 966 Active: 3237152 kB 967 Inactive: 7586256 kB 968 Active(anon): 94064 kB 969 Inactive(anon): 4570616 kB 970 Active(file): 3143088 kB 971 Inactive(file): 3015640 kB 972 Unevictable: 0 kB 973 Mlocked: 0 kB 974 SwapTotal: 0 kB 975 SwapFree: 0 kB 976 Zswap: 1904 kB 977 Zswapped: 7792 kB 978 Dirty: 12 kB 979 Writeback: 0 kB 980 AnonPages: 4654780 kB 981 Mapped: 266244 kB 982 Shmem: 9976 kB 983 KReclaimable: 517708 kB 984 Slab: 660044 kB 985 SReclaimable: 517708 kB 986 SUnreclaim: 142336 kB 987 KernelStack: 11168 kB 988 PageTables: 20540 kB 989 SecPageTables: 0 kB 990 NFS_Unstable: 0 kB 991 Bounce: 0 kB 992 WritebackTmp: 0 kB 993 CommitLimit: 16429408 kB 994 Committed_AS: 7715148 kB 995 VmallocTotal: 34359738367 kB 996 VmallocUsed: 40444 kB 997 VmallocChunk: 0 kB 998 Percpu: 29312 kB 999 HardwareCorrupted: 0 kB 1000 AnonHugePages: 4149248 kB 1001 ShmemHugePages: 0 kB 1002 ShmemPmdMapped: 0 kB 1003 FileHugePages: 0 kB 1004 FilePmdMapped: 0 kB 1005 CmaTotal: 0 kB 1006 CmaFree: 0 kB 1007 HugePages_Total: 0 1008 HugePages_Free: 0 1009 HugePages_Rsvd: 0 1010 HugePages_Surp: 0 1011 Hugepagesize: 2048 kB 1012 Hugetlb: 0 kB 1013 DirectMap4k: 401152 kB 1014 DirectMap2M: 10008576 kB 1015 DirectMap1G: 24117248 kB 1016 1017MemTotal 1018 Total usable RAM (i.e. physical RAM minus a few reserved 1019 bits and the kernel binary code) 1020MemFree 1021 Total free RAM. On highmem systems, the sum of LowFree+HighFree 1022MemAvailable 1023 An estimate of how much memory is available for starting new 1024 applications, without swapping. Calculated from MemFree, 1025 SReclaimable, the size of the file LRU lists, and the low 1026 watermarks in each zone. 1027 The estimate takes into account that the system needs some 1028 page cache to function well, and that not all reclaimable 1029 slab will be reclaimable, due to items being in use. The 1030 impact of those factors will vary from system to system. 1031Buffers 1032 Relatively temporary storage for raw disk blocks 1033 shouldn't get tremendously large (20MB or so) 1034Cached 1035 In-memory cache for files read from the disk (the 1036 pagecache) as well as tmpfs & shmem. 1037 Doesn't include SwapCached. 1038SwapCached 1039 Memory that once was swapped out, is swapped back in but 1040 still also is in the swapfile (if memory is needed it 1041 doesn't need to be swapped out AGAIN because it is already 1042 in the swapfile. This saves I/O) 1043Active 1044 Memory that has been used more recently and usually not 1045 reclaimed unless absolutely necessary. 1046Inactive 1047 Memory which has been less recently used. It is more 1048 eligible to be reclaimed for other purposes 1049Unevictable 1050 Memory allocated for userspace which cannot be reclaimed, such 1051 as mlocked pages, ramfs backing pages, secret memfd pages etc. 1052Mlocked 1053 Memory locked with mlock(). 1054HighTotal, HighFree 1055 Highmem is all memory above ~860MB of physical memory. 1056 Highmem areas are for use by userspace programs, or 1057 for the pagecache. The kernel must use tricks to access 1058 this memory, making it slower to access than lowmem. 1059LowTotal, LowFree 1060 Lowmem is memory which can be used for everything that 1061 highmem can be used for, but it is also available for the 1062 kernel's use for its own data structures. Among many 1063 other things, it is where everything from the Slab is 1064 allocated. Bad things happen when you're out of lowmem. 1065SwapTotal 1066 total amount of swap space available 1067SwapFree 1068 Memory which has been evicted from RAM, and is temporarily 1069 on the disk 1070Zswap 1071 Memory consumed by the zswap backend (compressed size) 1072Zswapped 1073 Amount of anonymous memory stored in zswap (original size) 1074Dirty 1075 Memory which is waiting to get written back to the disk 1076Writeback 1077 Memory which is actively being written back to the disk 1078AnonPages 1079 Non-file backed pages mapped into userspace page tables 1080Mapped 1081 files which have been mmaped, such as libraries 1082Shmem 1083 Total memory used by shared memory (shmem) and tmpfs 1084KReclaimable 1085 Kernel allocations that the kernel will attempt to reclaim 1086 under memory pressure. Includes SReclaimable (below), and other 1087 direct allocations with a shrinker. 1088Slab 1089 in-kernel data structures cache 1090SReclaimable 1091 Part of Slab, that might be reclaimed, such as caches 1092SUnreclaim 1093 Part of Slab, that cannot be reclaimed on memory pressure 1094KernelStack 1095 Memory consumed by the kernel stacks of all tasks 1096PageTables 1097 Memory consumed by userspace page tables 1098SecPageTables 1099 Memory consumed by secondary page tables, this currently 1100 currently includes KVM mmu allocations on x86 and arm64. 1101NFS_Unstable 1102 Always zero. Previous counted pages which had been written to 1103 the server, but has not been committed to stable storage. 1104Bounce 1105 Memory used for block device "bounce buffers" 1106WritebackTmp 1107 Memory used by FUSE for temporary writeback buffers 1108CommitLimit 1109 Based on the overcommit ratio ('vm.overcommit_ratio'), 1110 this is the total amount of memory currently available to 1111 be allocated on the system. This limit is only adhered to 1112 if strict overcommit accounting is enabled (mode 2 in 1113 'vm.overcommit_memory'). 1114 1115 The CommitLimit is calculated with the following formula:: 1116 1117 CommitLimit = ([total RAM pages] - [total huge TLB pages]) * 1118 overcommit_ratio / 100 + [total swap pages] 1119 1120 For example, on a system with 1G of physical RAM and 7G 1121 of swap with a `vm.overcommit_ratio` of 30 it would 1122 yield a CommitLimit of 7.3G. 1123 1124 For more details, see the memory overcommit documentation 1125 in mm/overcommit-accounting. 1126Committed_AS 1127 The amount of memory presently allocated on the system. 1128 The committed memory is a sum of all of the memory which 1129 has been allocated by processes, even if it has not been 1130 "used" by them as of yet. A process which malloc()'s 1G 1131 of memory, but only touches 300M of it will show up as 1132 using 1G. This 1G is memory which has been "committed" to 1133 by the VM and can be used at any time by the allocating 1134 application. With strict overcommit enabled on the system 1135 (mode 2 in 'vm.overcommit_memory'), allocations which would 1136 exceed the CommitLimit (detailed above) will not be permitted. 1137 This is useful if one needs to guarantee that processes will 1138 not fail due to lack of memory once that memory has been 1139 successfully allocated. 1140VmallocTotal 1141 total size of vmalloc virtual address space 1142VmallocUsed 1143 amount of vmalloc area which is used 1144VmallocChunk 1145 largest contiguous block of vmalloc area which is free 1146Percpu 1147 Memory allocated to the percpu allocator used to back percpu 1148 allocations. This stat excludes the cost of metadata. 1149HardwareCorrupted 1150 The amount of RAM/memory in KB, the kernel identifies as 1151 corrupted. 1152AnonHugePages 1153 Non-file backed huge pages mapped into userspace page tables 1154ShmemHugePages 1155 Memory used by shared memory (shmem) and tmpfs allocated 1156 with huge pages 1157ShmemPmdMapped 1158 Shared memory mapped into userspace with huge pages 1159FileHugePages 1160 Memory used for filesystem data (page cache) allocated 1161 with huge pages 1162FilePmdMapped 1163 Page cache mapped into userspace with huge pages 1164CmaTotal 1165 Memory reserved for the Contiguous Memory Allocator (CMA) 1166CmaFree 1167 Free remaining memory in the CMA reserves 1168HugePages_Total, HugePages_Free, HugePages_Rsvd, HugePages_Surp, Hugepagesize, Hugetlb 1169 See Documentation/admin-guide/mm/hugetlbpage.rst. 1170DirectMap4k, DirectMap2M, DirectMap1G 1171 Breakdown of page table sizes used in the kernel's 1172 identity mapping of RAM 1173 1174vmallocinfo 1175~~~~~~~~~~~ 1176 1177Provides information about vmalloced/vmaped areas. One line per area, 1178containing the virtual address range of the area, size in bytes, 1179caller information of the creator, and optional information depending 1180on the kind of area: 1181 1182 ========== =================================================== 1183 pages=nr number of pages 1184 phys=addr if a physical address was specified 1185 ioremap I/O mapping (ioremap() and friends) 1186 vmalloc vmalloc() area 1187 vmap vmap()ed pages 1188 user VM_USERMAP area 1189 vpages buffer for pages pointers was vmalloced (huge area) 1190 N<node>=nr (Only on NUMA kernels) 1191 Number of pages allocated on memory node <node> 1192 ========== =================================================== 1193 1194:: 1195 1196 > cat /proc/vmallocinfo 1197 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ... 1198 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128 1199 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ... 1200 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64 1201 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f... 1202 phys=7fee8000 ioremap 1203 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f... 1204 phys=7fee7000 ioremap 1205 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210 1206 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ... 1207 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3 1208 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ... 1209 pages=2 vmalloc N1=2 1210 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ... 1211 /0x130 [x_tables] pages=4 vmalloc N0=4 1212 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ... 1213 pages=14 vmalloc N2=14 1214 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ... 1215 pages=4 vmalloc N1=4 1216 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ... 1217 pages=2 vmalloc N1=2 1218 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ... 1219 pages=10 vmalloc N0=10 1220 1221 1222softirqs 1223~~~~~~~~ 1224 1225Provides counts of softirq handlers serviced since boot time, for each CPU. 1226 1227:: 1228 1229 > cat /proc/softirqs 1230 CPU0 CPU1 CPU2 CPU3 1231 HI: 0 0 0 0 1232 TIMER: 27166 27120 27097 27034 1233 NET_TX: 0 0 0 17 1234 NET_RX: 42 0 0 39 1235 BLOCK: 0 0 107 1121 1236 TASKLET: 0 0 0 290 1237 SCHED: 27035 26983 26971 26746 1238 HRTIMER: 0 0 0 0 1239 RCU: 1678 1769 2178 2250 1240 12411.3 Networking info in /proc/net 1242-------------------------------- 1243 1244The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the 1245additional values you get for IP version 6 if you configure the kernel to 1246support this. Table 1-9 lists the files and their meaning. 1247 1248 1249.. table:: Table 1-8: IPv6 info in /proc/net 1250 1251 ========== ===================================================== 1252 File Content 1253 ========== ===================================================== 1254 udp6 UDP sockets (IPv6) 1255 tcp6 TCP sockets (IPv6) 1256 raw6 Raw device statistics (IPv6) 1257 igmp6 IP multicast addresses, which this host joined (IPv6) 1258 if_inet6 List of IPv6 interface addresses 1259 ipv6_route Kernel routing table for IPv6 1260 rt6_stats Global IPv6 routing tables statistics 1261 sockstat6 Socket statistics (IPv6) 1262 snmp6 Snmp data (IPv6) 1263 ========== ===================================================== 1264 1265.. table:: Table 1-9: Network info in /proc/net 1266 1267 ============= ================================================================ 1268 File Content 1269 ============= ================================================================ 1270 arp Kernel ARP table 1271 dev network devices with statistics 1272 dev_mcast the Layer2 multicast groups a device is listening too 1273 (interface index, label, number of references, number of bound 1274 addresses). 1275 dev_stat network device status 1276 ip_fwchains Firewall chain linkage 1277 ip_fwnames Firewall chain names 1278 ip_masq Directory containing the masquerading tables 1279 ip_masquerade Major masquerading table 1280 netstat Network statistics 1281 raw raw device statistics 1282 route Kernel routing table 1283 rpc Directory containing rpc info 1284 rt_cache Routing cache 1285 snmp SNMP data 1286 sockstat Socket statistics 1287 tcp TCP sockets 1288 udp UDP sockets 1289 unix UNIX domain sockets 1290 wireless Wireless interface data (Wavelan etc) 1291 igmp IP multicast addresses, which this host joined 1292 psched Global packet scheduler parameters. 1293 netlink List of PF_NETLINK sockets 1294 ip_mr_vifs List of multicast virtual interfaces 1295 ip_mr_cache List of multicast routing cache 1296 ============= ================================================================ 1297 1298You can use this information to see which network devices are available in 1299your system and how much traffic was routed over those devices:: 1300 1301 > cat /proc/net/dev 1302 Inter-|Receive |[... 1303 face |bytes packets errs drop fifo frame compressed multicast|[... 1304 lo: 908188 5596 0 0 0 0 0 0 [... 1305 ppp0:15475140 20721 410 0 0 410 0 0 [... 1306 eth0: 614530 7085 0 0 0 0 0 1 [... 1307 1308 ...] Transmit 1309 ...] bytes packets errs drop fifo colls carrier compressed 1310 ...] 908188 5596 0 0 0 0 0 0 1311 ...] 1375103 17405 0 0 0 0 0 0 1312 ...] 1703981 5535 0 0 0 3 0 0 1313 1314In addition, each Channel Bond interface has its own directory. For 1315example, the bond0 device will have a directory called /proc/net/bond0/. 1316It will contain information that is specific to that bond, such as the 1317current slaves of the bond, the link status of the slaves, and how 1318many times the slaves link has failed. 1319 13201.4 SCSI info 1321------------- 1322 1323If you have a SCSI host adapter in your system, you'll find a subdirectory 1324named after the driver for this adapter in /proc/scsi. You'll also see a list 1325of all recognized SCSI devices in /proc/scsi:: 1326 1327 >cat /proc/scsi/scsi 1328 Attached devices: 1329 Host: scsi0 Channel: 00 Id: 00 Lun: 00 1330 Vendor: IBM Model: DGHS09U Rev: 03E0 1331 Type: Direct-Access ANSI SCSI revision: 03 1332 Host: scsi0 Channel: 00 Id: 06 Lun: 00 1333 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04 1334 Type: CD-ROM ANSI SCSI revision: 02 1335 1336 1337The directory named after the driver has one file for each adapter found in 1338the system. These files contain information about the controller, including 1339the used IRQ and the IO address range. The amount of information shown is 1340dependent on the adapter you use. The example shows the output for an Adaptec 1341AHA-2940 SCSI adapter:: 1342 1343 > cat /proc/scsi/aic7xxx/0 1344 1345 Adaptec AIC7xxx driver version: 5.1.19/3.2.4 1346 Compile Options: 1347 TCQ Enabled By Default : Disabled 1348 AIC7XXX_PROC_STATS : Disabled 1349 AIC7XXX_RESET_DELAY : 5 1350 Adapter Configuration: 1351 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter 1352 Ultra Wide Controller 1353 PCI MMAPed I/O Base: 0xeb001000 1354 Adapter SEEPROM Config: SEEPROM found and used. 1355 Adaptec SCSI BIOS: Enabled 1356 IRQ: 10 1357 SCBs: Active 0, Max Active 2, 1358 Allocated 15, HW 16, Page 255 1359 Interrupts: 160328 1360 BIOS Control Word: 0x18b6 1361 Adapter Control Word: 0x005b 1362 Extended Translation: Enabled 1363 Disconnect Enable Flags: 0xffff 1364 Ultra Enable Flags: 0x0001 1365 Tag Queue Enable Flags: 0x0000 1366 Ordered Queue Tag Flags: 0x0000 1367 Default Tag Queue Depth: 8 1368 Tagged Queue By Device array for aic7xxx host instance 0: 1369 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255} 1370 Actual queue depth per device for aic7xxx host instance 0: 1371 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} 1372 Statistics: 1373 (scsi0:0:0:0) 1374 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8 1375 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0) 1376 Total transfers 160151 (74577 reads and 85574 writes) 1377 (scsi0:0:6:0) 1378 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15 1379 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0) 1380 Total transfers 0 (0 reads and 0 writes) 1381 1382 13831.5 Parallel port info in /proc/parport 1384--------------------------------------- 1385 1386The directory /proc/parport contains information about the parallel ports of 1387your system. It has one subdirectory for each port, named after the port 1388number (0,1,2,...). 1389 1390These directories contain the four files shown in Table 1-10. 1391 1392 1393.. table:: Table 1-10: Files in /proc/parport 1394 1395 ========= ==================================================================== 1396 File Content 1397 ========= ==================================================================== 1398 autoprobe Any IEEE-1284 device ID information that has been acquired. 1399 devices list of the device drivers using that port. A + will appear by the 1400 name of the device currently using the port (it might not appear 1401 against any). 1402 hardware Parallel port's base address, IRQ line and DMA channel. 1403 irq IRQ that parport is using for that port. This is in a separate 1404 file to allow you to alter it by writing a new value in (IRQ 1405 number or none). 1406 ========= ==================================================================== 1407 14081.6 TTY info in /proc/tty 1409------------------------- 1410 1411Information about the available and actually used tty's can be found in the 1412directory /proc/tty. You'll find entries for drivers and line disciplines in 1413this directory, as shown in Table 1-11. 1414 1415 1416.. table:: Table 1-11: Files in /proc/tty 1417 1418 ============= ============================================== 1419 File Content 1420 ============= ============================================== 1421 drivers list of drivers and their usage 1422 ldiscs registered line disciplines 1423 driver/serial usage statistic and status of single tty lines 1424 ============= ============================================== 1425 1426To see which tty's are currently in use, you can simply look into the file 1427/proc/tty/drivers:: 1428 1429 > cat /proc/tty/drivers 1430 pty_slave /dev/pts 136 0-255 pty:slave 1431 pty_master /dev/ptm 128 0-255 pty:master 1432 pty_slave /dev/ttyp 3 0-255 pty:slave 1433 pty_master /dev/pty 2 0-255 pty:master 1434 serial /dev/cua 5 64-67 serial:callout 1435 serial /dev/ttyS 4 64-67 serial 1436 /dev/tty0 /dev/tty0 4 0 system:vtmaster 1437 /dev/ptmx /dev/ptmx 5 2 system 1438 /dev/console /dev/console 5 1 system:console 1439 /dev/tty /dev/tty 5 0 system:/dev/tty 1440 unknown /dev/tty 4 1-63 console 1441 1442 14431.7 Miscellaneous kernel statistics in /proc/stat 1444------------------------------------------------- 1445 1446Various pieces of information about kernel activity are available in the 1447/proc/stat file. All of the numbers reported in this file are aggregates 1448since the system first booted. For a quick look, simply cat the file:: 1449 1450 > cat /proc/stat 1451 cpu 2255 34 2290 22625563 6290 127 456 0 0 0 1452 cpu0 1132 34 1441 11311718 3675 127 438 0 0 0 1453 cpu1 1123 0 849 11313845 2614 0 18 0 0 0 1454 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...] 1455 ctxt 1990473 1456 btime 1062191376 1457 processes 2915 1458 procs_running 1 1459 procs_blocked 0 1460 softirq 183433 0 21755 12 39 1137 231 21459 2263 1461 1462The very first "cpu" line aggregates the numbers in all of the other "cpuN" 1463lines. These numbers identify the amount of time the CPU has spent performing 1464different kinds of work. Time units are in USER_HZ (typically hundredths of a 1465second). The meanings of the columns are as follows, from left to right: 1466 1467- user: normal processes executing in user mode 1468- nice: niced processes executing in user mode 1469- system: processes executing in kernel mode 1470- idle: twiddling thumbs 1471- iowait: In a word, iowait stands for waiting for I/O to complete. But there 1472 are several problems: 1473 1474 1. CPU will not wait for I/O to complete, iowait is the time that a task is 1475 waiting for I/O to complete. When CPU goes into idle state for 1476 outstanding task I/O, another task will be scheduled on this CPU. 1477 2. In a multi-core CPU, the task waiting for I/O to complete is not running 1478 on any CPU, so the iowait of each CPU is difficult to calculate. 1479 3. The value of iowait field in /proc/stat will decrease in certain 1480 conditions. 1481 1482 So, the iowait is not reliable by reading from /proc/stat. 1483- irq: servicing interrupts 1484- softirq: servicing softirqs 1485- steal: involuntary wait 1486- guest: running a normal guest 1487- guest_nice: running a niced guest 1488 1489The "intr" line gives counts of interrupts serviced since boot time, for each 1490of the possible system interrupts. The first column is the total of all 1491interrupts serviced including unnumbered architecture specific interrupts; 1492each subsequent column is the total for that particular numbered interrupt. 1493Unnumbered interrupts are not shown, only summed into the total. 1494 1495The "ctxt" line gives the total number of context switches across all CPUs. 1496 1497The "btime" line gives the time at which the system booted, in seconds since 1498the Unix epoch. 1499 1500The "processes" line gives the number of processes and threads created, which 1501includes (but is not limited to) those created by calls to the fork() and 1502clone() system calls. 1503 1504The "procs_running" line gives the total number of threads that are 1505running or ready to run (i.e., the total number of runnable threads). 1506 1507The "procs_blocked" line gives the number of processes currently blocked, 1508waiting for I/O to complete. 1509 1510The "softirq" line gives counts of softirqs serviced since boot time, for each 1511of the possible system softirqs. The first column is the total of all 1512softirqs serviced; each subsequent column is the total for that particular 1513softirq. 1514 1515 15161.8 Ext4 file system parameters 1517------------------------------- 1518 1519Information about mounted ext4 file systems can be found in 1520/proc/fs/ext4. Each mounted filesystem will have a directory in 1521/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or 1522/proc/fs/ext4/dm-0). The files in each per-device directory are shown 1523in Table 1-12, below. 1524 1525.. table:: Table 1-12: Files in /proc/fs/ext4/<devname> 1526 1527 ============== ========================================================== 1528 File Content 1529 mb_groups details of multiblock allocator buddy cache of free blocks 1530 ============== ========================================================== 1531 15321.9 /proc/consoles 1533------------------- 1534Shows registered system console lines. 1535 1536To see which character device lines are currently used for the system console 1537/dev/console, you may simply look into the file /proc/consoles:: 1538 1539 > cat /proc/consoles 1540 tty0 -WU (ECp) 4:7 1541 ttyS0 -W- (Ep) 4:64 1542 1543The columns are: 1544 1545+--------------------+-------------------------------------------------------+ 1546| device | name of the device | 1547+====================+=======================================================+ 1548| operations | * R = can do read operations | 1549| | * W = can do write operations | 1550| | * U = can do unblank | 1551+--------------------+-------------------------------------------------------+ 1552| flags | * E = it is enabled | 1553| | * C = it is preferred console | 1554| | * B = it is primary boot console | 1555| | * p = it is used for printk buffer | 1556| | * b = it is not a TTY but a Braille device | 1557| | * a = it is safe to use when cpu is offline | 1558+--------------------+-------------------------------------------------------+ 1559| major:minor | major and minor number of the device separated by a | 1560| | colon | 1561+--------------------+-------------------------------------------------------+ 1562 1563Summary 1564------- 1565 1566The /proc file system serves information about the running system. It not only 1567allows access to process data but also allows you to request the kernel status 1568by reading files in the hierarchy. 1569 1570The directory structure of /proc reflects the types of information and makes 1571it easy, if not obvious, where to look for specific data. 1572 1573Chapter 2: Modifying System Parameters 1574====================================== 1575 1576In This Chapter 1577--------------- 1578 1579* Modifying kernel parameters by writing into files found in /proc/sys 1580* Exploring the files which modify certain parameters 1581* Review of the /proc/sys file tree 1582 1583------------------------------------------------------------------------------ 1584 1585A very interesting part of /proc is the directory /proc/sys. This is not only 1586a source of information, it also allows you to change parameters within the 1587kernel. Be very careful when attempting this. You can optimize your system, 1588but you can also cause it to crash. Never alter kernel parameters on a 1589production system. Set up a development machine and test to make sure that 1590everything works the way you want it to. You may have no alternative but to 1591reboot the machine once an error has been made. 1592 1593To change a value, simply echo the new value into the file. 1594You need to be root to do this. You can create your own boot script 1595to perform this every time your system boots. 1596 1597The files in /proc/sys can be used to fine tune and monitor miscellaneous and 1598general things in the operation of the Linux kernel. Since some of the files 1599can inadvertently disrupt your system, it is advisable to read both 1600documentation and source before actually making adjustments. In any case, be 1601very careful when writing to any of these files. The entries in /proc may 1602change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt 1603review the kernel documentation in the directory /usr/src/linux/Documentation. 1604This chapter is heavily based on the documentation included in the pre 2.2 1605kernels, and became part of it in version 2.2.1 of the Linux kernel. 1606 1607Please see: Documentation/admin-guide/sysctl/ directory for descriptions of these 1608entries. 1609 1610Summary 1611------- 1612 1613Certain aspects of kernel behavior can be modified at runtime, without the 1614need to recompile the kernel, or even to reboot the system. The files in the 1615/proc/sys tree can not only be read, but also modified. You can use the echo 1616command to write value into these files, thereby changing the default settings 1617of the kernel. 1618 1619 1620Chapter 3: Per-process Parameters 1621================================= 1622 16233.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score 1624-------------------------------------------------------------------------------- 1625 1626These files can be used to adjust the badness heuristic used to select which 1627process gets killed in out of memory (oom) conditions. 1628 1629The badness heuristic assigns a value to each candidate task ranging from 0 1630(never kill) to 1000 (always kill) to determine which process is targeted. The 1631units are roughly a proportion along that range of allowed memory the process 1632may allocate from based on an estimation of its current memory and swap use. 1633For example, if a task is using all allowed memory, its badness score will be 16341000. If it is using half of its allowed memory, its score will be 500. 1635 1636The amount of "allowed" memory depends on the context in which the oom killer 1637was called. If it is due to the memory assigned to the allocating task's cpuset 1638being exhausted, the allowed memory represents the set of mems assigned to that 1639cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed 1640memory represents the set of mempolicy nodes. If it is due to a memory 1641limit (or swap limit) being reached, the allowed memory is that configured 1642limit. Finally, if it is due to the entire system being out of memory, the 1643allowed memory represents all allocatable resources. 1644 1645The value of /proc/<pid>/oom_score_adj is added to the badness score before it 1646is used to determine which task to kill. Acceptable values range from -1000 1647(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to 1648polarize the preference for oom killing either by always preferring a certain 1649task or completely disabling it. The lowest possible value, -1000, is 1650equivalent to disabling oom killing entirely for that task since it will always 1651report a badness score of 0. 1652 1653Consequently, it is very simple for userspace to define the amount of memory to 1654consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for 1655example, is roughly equivalent to allowing the remainder of tasks sharing the 1656same system, cpuset, mempolicy, or memory controller resources to use at least 165750% more memory. A value of -500, on the other hand, would be roughly 1658equivalent to discounting 50% of the task's allowed memory from being considered 1659as scoring against the task. 1660 1661For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also 1662be used to tune the badness score. Its acceptable values range from -16 1663(OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17 1664(OOM_DISABLE) to disable oom killing entirely for that task. Its value is 1665scaled linearly with /proc/<pid>/oom_score_adj. 1666 1667The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last 1668value set by a CAP_SYS_RESOURCE process. To reduce the value any lower 1669requires CAP_SYS_RESOURCE. 1670 1671 16723.2 /proc/<pid>/oom_score - Display current oom-killer score 1673------------------------------------------------------------- 1674 1675This file can be used to check the current score used by the oom-killer for 1676any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which 1677process should be killed in an out-of-memory situation. 1678 1679Please note that the exported value includes oom_score_adj so it is 1680effectively in range [0,2000]. 1681 1682 16833.3 /proc/<pid>/io - Display the IO accounting fields 1684------------------------------------------------------- 1685 1686This file contains IO statistics for each running process. 1687 1688Example 1689~~~~~~~ 1690 1691:: 1692 1693 test:/tmp # dd if=/dev/zero of=/tmp/test.dat & 1694 [1] 3828 1695 1696 test:/tmp # cat /proc/3828/io 1697 rchar: 323934931 1698 wchar: 323929600 1699 syscr: 632687 1700 syscw: 632675 1701 read_bytes: 0 1702 write_bytes: 323932160 1703 cancelled_write_bytes: 0 1704 1705 1706Description 1707~~~~~~~~~~~ 1708 1709rchar 1710^^^^^ 1711 1712I/O counter: chars read 1713The number of bytes which this task has caused to be read from storage. This 1714is simply the sum of bytes which this process passed to read() and pread(). 1715It includes things like tty IO and it is unaffected by whether or not actual 1716physical disk IO was required (the read might have been satisfied from 1717pagecache). 1718 1719 1720wchar 1721^^^^^ 1722 1723I/O counter: chars written 1724The number of bytes which this task has caused, or shall cause to be written 1725to disk. Similar caveats apply here as with rchar. 1726 1727 1728syscr 1729^^^^^ 1730 1731I/O counter: read syscalls 1732Attempt to count the number of read I/O operations, i.e. syscalls like read() 1733and pread(). 1734 1735 1736syscw 1737^^^^^ 1738 1739I/O counter: write syscalls 1740Attempt to count the number of write I/O operations, i.e. syscalls like 1741write() and pwrite(). 1742 1743 1744read_bytes 1745^^^^^^^^^^ 1746 1747I/O counter: bytes read 1748Attempt to count the number of bytes which this process really did cause to 1749be fetched from the storage layer. Done at the submit_bio() level, so it is 1750accurate for block-backed filesystems. <please add status regarding NFS and 1751CIFS at a later time> 1752 1753 1754write_bytes 1755^^^^^^^^^^^ 1756 1757I/O counter: bytes written 1758Attempt to count the number of bytes which this process caused to be sent to 1759the storage layer. This is done at page-dirtying time. 1760 1761 1762cancelled_write_bytes 1763^^^^^^^^^^^^^^^^^^^^^ 1764 1765The big inaccuracy here is truncate. If a process writes 1MB to a file and 1766then deletes the file, it will in fact perform no writeout. But it will have 1767been accounted as having caused 1MB of write. 1768In other words: The number of bytes which this process caused to not happen, 1769by truncating pagecache. A task can cause "negative" IO too. If this task 1770truncates some dirty pagecache, some IO which another task has been accounted 1771for (in its write_bytes) will not be happening. We _could_ just subtract that 1772from the truncating task's write_bytes, but there is information loss in doing 1773that. 1774 1775 1776.. Note:: 1777 1778 At its current implementation state, this is a bit racy on 32-bit machines: 1779 if process A reads process B's /proc/pid/io while process B is updating one 1780 of those 64-bit counters, process A could see an intermediate result. 1781 1782 1783More information about this can be found within the taskstats documentation in 1784Documentation/accounting. 1785 17863.4 /proc/<pid>/coredump_filter - Core dump filtering settings 1787--------------------------------------------------------------- 1788When a process is dumped, all anonymous memory is written to a core file as 1789long as the size of the core file isn't limited. But sometimes we don't want 1790to dump some memory segments, for example, huge shared memory or DAX. 1791Conversely, sometimes we want to save file-backed memory segments into a core 1792file, not only the individual files. 1793 1794/proc/<pid>/coredump_filter allows you to customize which memory segments 1795will be dumped when the <pid> process is dumped. coredump_filter is a bitmask 1796of memory types. If a bit of the bitmask is set, memory segments of the 1797corresponding memory type are dumped, otherwise they are not dumped. 1798 1799The following 9 memory types are supported: 1800 1801 - (bit 0) anonymous private memory 1802 - (bit 1) anonymous shared memory 1803 - (bit 2) file-backed private memory 1804 - (bit 3) file-backed shared memory 1805 - (bit 4) ELF header pages in file-backed private memory areas (it is 1806 effective only if the bit 2 is cleared) 1807 - (bit 5) hugetlb private memory 1808 - (bit 6) hugetlb shared memory 1809 - (bit 7) DAX private memory 1810 - (bit 8) DAX shared memory 1811 1812 Note that MMIO pages such as frame buffer are never dumped and vDSO pages 1813 are always dumped regardless of the bitmask status. 1814 1815 Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is 1816 only affected by bit 5-6, and DAX is only affected by bits 7-8. 1817 1818The default value of coredump_filter is 0x33; this means all anonymous memory 1819segments, ELF header pages and hugetlb private memory are dumped. 1820 1821If you don't want to dump all shared memory segments attached to pid 1234, 1822write 0x31 to the process's proc file:: 1823 1824 $ echo 0x31 > /proc/1234/coredump_filter 1825 1826When a new process is created, the process inherits the bitmask status from its 1827parent. It is useful to set up coredump_filter before the program runs. 1828For example:: 1829 1830 $ echo 0x7 > /proc/self/coredump_filter 1831 $ ./some_program 1832 18333.5 /proc/<pid>/mountinfo - Information about mounts 1834-------------------------------------------------------- 1835 1836This file contains lines of the form:: 1837 1838 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue 1839 (1)(2)(3) (4) (5) (6) (n…m) (m+1)(m+2) (m+3) (m+4) 1840 1841 (1) mount ID: unique identifier of the mount (may be reused after umount) 1842 (2) parent ID: ID of parent (or of self for the top of the mount tree) 1843 (3) major:minor: value of st_dev for files on filesystem 1844 (4) root: root of the mount within the filesystem 1845 (5) mount point: mount point relative to the process's root 1846 (6) mount options: per mount options 1847 (n…m) optional fields: zero or more fields of the form "tag[:value]" 1848 (m+1) separator: marks the end of the optional fields 1849 (m+2) filesystem type: name of filesystem of the form "type[.subtype]" 1850 (m+3) mount source: filesystem specific information or "none" 1851 (m+4) super options: per super block options 1852 1853Parsers should ignore all unrecognised optional fields. Currently the 1854possible optional fields are: 1855 1856================ ============================================================== 1857shared:X mount is shared in peer group X 1858master:X mount is slave to peer group X 1859propagate_from:X mount is slave and receives propagation from peer group X [#]_ 1860unbindable mount is unbindable 1861================ ============================================================== 1862 1863.. [#] X is the closest dominant peer group under the process's root. If 1864 X is the immediate master of the mount, or if there's no dominant peer 1865 group under the same root, then only the "master:X" field is present 1866 and not the "propagate_from:X" field. 1867 1868For more information on mount propagation see: 1869 1870 Documentation/filesystems/sharedsubtree.rst 1871 1872 18733.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 1874-------------------------------------------------------- 1875These files provide a method to access a task's comm value. It also allows for 1876a task to set its own or one of its thread siblings comm value. The comm value 1877is limited in size compared to the cmdline value, so writing anything longer 1878then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated 1879comm value. 1880 1881 18823.7 /proc/<pid>/task/<tid>/children - Information about task children 1883------------------------------------------------------------------------- 1884This file provides a fast way to retrieve first level children pids 1885of a task pointed by <pid>/<tid> pair. The format is a space separated 1886stream of pids. 1887 1888Note the "first level" here -- if a child has its own children they will 1889not be listed here; one needs to read /proc/<children-pid>/task/<tid>/children 1890to obtain the descendants. 1891 1892Since this interface is intended to be fast and cheap it doesn't 1893guarantee to provide precise results and some children might be 1894skipped, especially if they've exited right after we printed their 1895pids, so one needs to either stop or freeze processes being inspected 1896if precise results are needed. 1897 1898 18993.8 /proc/<pid>/fdinfo/<fd> - Information about opened file 1900--------------------------------------------------------------- 1901This file provides information associated with an opened file. The regular 1902files have at least four fields -- 'pos', 'flags', 'mnt_id' and 'ino'. 1903The 'pos' represents the current offset of the opened file in decimal 1904form [see lseek(2) for details], 'flags' denotes the octal O_xxx mask the 1905file has been created with [see open(2) for details] and 'mnt_id' represents 1906mount ID of the file system containing the opened file [see 3.5 1907/proc/<pid>/mountinfo for details]. 'ino' represents the inode number of 1908the file. 1909 1910A typical output is:: 1911 1912 pos: 0 1913 flags: 0100002 1914 mnt_id: 19 1915 ino: 63107 1916 1917All locks associated with a file descriptor are shown in its fdinfo too:: 1918 1919 lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF 1920 1921The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags 1922pair provide additional information particular to the objects they represent. 1923 1924Eventfd files 1925~~~~~~~~~~~~~ 1926 1927:: 1928 1929 pos: 0 1930 flags: 04002 1931 mnt_id: 9 1932 ino: 63107 1933 eventfd-count: 5a 1934 1935where 'eventfd-count' is hex value of a counter. 1936 1937Signalfd files 1938~~~~~~~~~~~~~~ 1939 1940:: 1941 1942 pos: 0 1943 flags: 04002 1944 mnt_id: 9 1945 ino: 63107 1946 sigmask: 0000000000000200 1947 1948where 'sigmask' is hex value of the signal mask associated 1949with a file. 1950 1951Epoll files 1952~~~~~~~~~~~ 1953 1954:: 1955 1956 pos: 0 1957 flags: 02 1958 mnt_id: 9 1959 ino: 63107 1960 tfd: 5 events: 1d data: ffffffffffffffff pos:0 ino:61af sdev:7 1961 1962where 'tfd' is a target file descriptor number in decimal form, 1963'events' is events mask being watched and the 'data' is data 1964associated with a target [see epoll(7) for more details]. 1965 1966The 'pos' is current offset of the target file in decimal form 1967[see lseek(2)], 'ino' and 'sdev' are inode and device numbers 1968where target file resides, all in hex format. 1969 1970Fsnotify files 1971~~~~~~~~~~~~~~ 1972For inotify files the format is the following:: 1973 1974 pos: 0 1975 flags: 02000000 1976 mnt_id: 9 1977 ino: 63107 1978 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d 1979 1980where 'wd' is a watch descriptor in decimal form, i.e. a target file 1981descriptor number, 'ino' and 'sdev' are inode and device where the 1982target file resides and the 'mask' is the mask of events, all in hex 1983form [see inotify(7) for more details]. 1984 1985If the kernel was built with exportfs support, the path to the target 1986file is encoded as a file handle. The file handle is provided by three 1987fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex 1988format. 1989 1990If the kernel is built without exportfs support the file handle won't be 1991printed out. 1992 1993If there is no inotify mark attached yet the 'inotify' line will be omitted. 1994 1995For fanotify files the format is:: 1996 1997 pos: 0 1998 flags: 02 1999 mnt_id: 9 2000 ino: 63107 2001 fanotify flags:10 event-flags:0 2002 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003 2003 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4 2004 2005where fanotify 'flags' and 'event-flags' are values used in fanotify_init 2006call, 'mnt_id' is the mount point identifier, 'mflags' is the value of 2007flags associated with mark which are tracked separately from events 2008mask. 'ino' and 'sdev' are target inode and device, 'mask' is the events 2009mask and 'ignored_mask' is the mask of events which are to be ignored. 2010All are in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask' 2011provide information about flags and mask used in fanotify_mark 2012call [see fsnotify manpage for details]. 2013 2014While the first three lines are mandatory and always printed, the rest is 2015optional and may be omitted if no marks created yet. 2016 2017Timerfd files 2018~~~~~~~~~~~~~ 2019 2020:: 2021 2022 pos: 0 2023 flags: 02 2024 mnt_id: 9 2025 ino: 63107 2026 clockid: 0 2027 ticks: 0 2028 settime flags: 01 2029 it_value: (0, 49406829) 2030 it_interval: (1, 0) 2031 2032where 'clockid' is the clock type and 'ticks' is the number of the timer expirations 2033that have occurred [see timerfd_create(2) for details]. 'settime flags' are 2034flags in octal form been used to setup the timer [see timerfd_settime(2) for 2035details]. 'it_value' is remaining time until the timer expiration. 2036'it_interval' is the interval for the timer. Note the timer might be set up 2037with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value' 2038still exhibits timer's remaining time. 2039 2040DMA Buffer files 2041~~~~~~~~~~~~~~~~ 2042 2043:: 2044 2045 pos: 0 2046 flags: 04002 2047 mnt_id: 9 2048 ino: 63107 2049 size: 32768 2050 count: 2 2051 exp_name: system-heap 2052 2053where 'size' is the size of the DMA buffer in bytes. 'count' is the file count of 2054the DMA buffer file. 'exp_name' is the name of the DMA buffer exporter. 2055 20563.9 /proc/<pid>/map_files - Information about memory mapped files 2057--------------------------------------------------------------------- 2058This directory contains symbolic links which represent memory mapped files 2059the process is maintaining. Example output:: 2060 2061 | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so 2062 | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so 2063 | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so 2064 | ... 2065 | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1 2066 | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls 2067 2068The name of a link represents the virtual memory bounds of a mapping, i.e. 2069vm_area_struct::vm_start-vm_area_struct::vm_end. 2070 2071The main purpose of the map_files is to retrieve a set of memory mapped 2072files in a fast way instead of parsing /proc/<pid>/maps or 2073/proc/<pid>/smaps, both of which contain many more records. At the same 2074time one can open(2) mappings from the listings of two processes and 2075comparing their inode numbers to figure out which anonymous memory areas 2076are actually shared. 2077 20783.10 /proc/<pid>/timerslack_ns - Task timerslack value 2079--------------------------------------------------------- 2080This file provides the value of the task's timerslack value in nanoseconds. 2081This value specifies an amount of time that normal timers may be deferred 2082in order to coalesce timers and avoid unnecessary wakeups. 2083 2084This allows a task's interactivity vs power consumption tradeoff to be 2085adjusted. 2086 2087Writing 0 to the file will set the task's timerslack to the default value. 2088 2089Valid values are from 0 - ULLONG_MAX 2090 2091An application setting the value must have PTRACE_MODE_ATTACH_FSCREDS level 2092permissions on the task specified to change its timerslack_ns value. 2093 20943.11 /proc/<pid>/patch_state - Livepatch patch operation state 2095----------------------------------------------------------------- 2096When CONFIG_LIVEPATCH is enabled, this file displays the value of the 2097patch state for the task. 2098 2099A value of '-1' indicates that no patch is in transition. 2100 2101A value of '0' indicates that a patch is in transition and the task is 2102unpatched. If the patch is being enabled, then the task hasn't been 2103patched yet. If the patch is being disabled, then the task has already 2104been unpatched. 2105 2106A value of '1' indicates that a patch is in transition and the task is 2107patched. If the patch is being enabled, then the task has already been 2108patched. If the patch is being disabled, then the task hasn't been 2109unpatched yet. 2110 21113.12 /proc/<pid>/arch_status - task architecture specific status 2112------------------------------------------------------------------- 2113When CONFIG_PROC_PID_ARCH_STATUS is enabled, this file displays the 2114architecture specific status of the task. 2115 2116Example 2117~~~~~~~ 2118 2119:: 2120 2121 $ cat /proc/6753/arch_status 2122 AVX512_elapsed_ms: 8 2123 2124Description 2125~~~~~~~~~~~ 2126 2127x86 specific entries 2128~~~~~~~~~~~~~~~~~~~~~ 2129 2130AVX512_elapsed_ms 2131^^^^^^^^^^^^^^^^^^ 2132 2133 If AVX512 is supported on the machine, this entry shows the milliseconds 2134 elapsed since the last time AVX512 usage was recorded. The recording 2135 happens on a best effort basis when a task is scheduled out. This means 2136 that the value depends on two factors: 2137 2138 1) The time which the task spent on the CPU without being scheduled 2139 out. With CPU isolation and a single runnable task this can take 2140 several seconds. 2141 2142 2) The time since the task was scheduled out last. Depending on the 2143 reason for being scheduled out (time slice exhausted, syscall ...) 2144 this can be arbitrary long time. 2145 2146 As a consequence the value cannot be considered precise and authoritative 2147 information. The application which uses this information has to be aware 2148 of the overall scenario on the system in order to determine whether a 2149 task is a real AVX512 user or not. Precise information can be obtained 2150 with performance counters. 2151 2152 A special value of '-1' indicates that no AVX512 usage was recorded, thus 2153 the task is unlikely an AVX512 user, but depends on the workload and the 2154 scheduling scenario, it also could be a false negative mentioned above. 2155 21563.13 /proc/<pid>/fd - List of symlinks to open files 2157------------------------------------------------------- 2158This directory contains symbolic links which represent open files 2159the process is maintaining. Example output:: 2160 2161 lr-x------ 1 root root 64 Sep 20 17:53 0 -> /dev/null 2162 l-wx------ 1 root root 64 Sep 20 17:53 1 -> /dev/null 2163 lrwx------ 1 root root 64 Sep 20 17:53 10 -> 'socket:[12539]' 2164 lrwx------ 1 root root 64 Sep 20 17:53 11 -> 'socket:[12540]' 2165 lrwx------ 1 root root 64 Sep 20 17:53 12 -> 'socket:[12542]' 2166 2167The number of open files for the process is stored in 'size' member 2168of stat() output for /proc/<pid>/fd for fast access. 2169------------------------------------------------------- 2170 2171 2172Chapter 4: Configuring procfs 2173============================= 2174 21754.1 Mount options 2176--------------------- 2177 2178The following mount options are supported: 2179 2180 ========= ======================================================== 2181 hidepid= Set /proc/<pid>/ access mode. 2182 gid= Set the group authorized to learn processes information. 2183 subset= Show only the specified subset of procfs. 2184 ========= ======================================================== 2185 2186hidepid=off or hidepid=0 means classic mode - everybody may access all 2187/proc/<pid>/ directories (default). 2188 2189hidepid=noaccess or hidepid=1 means users may not access any /proc/<pid>/ 2190directories but their own. Sensitive files like cmdline, sched*, status are now 2191protected against other users. This makes it impossible to learn whether any 2192user runs specific program (given the program doesn't reveal itself by its 2193behaviour). As an additional bonus, as /proc/<pid>/cmdline is unaccessible for 2194other users, poorly written programs passing sensitive information via program 2195arguments are now protected against local eavesdroppers. 2196 2197hidepid=invisible or hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be 2198fully invisible to other users. It doesn't mean that it hides a fact whether a 2199process with a specific pid value exists (it can be learned by other means, e.g. 2200by "kill -0 $PID"), but it hides process' uid and gid, which may be learned by 2201stat()'ing /proc/<pid>/ otherwise. It greatly complicates an intruder's task of 2202gathering information about running processes, whether some daemon runs with 2203elevated privileges, whether other user runs some sensitive program, whether 2204other users run any program at all, etc. 2205 2206hidepid=ptraceable or hidepid=4 means that procfs should only contain 2207/proc/<pid>/ directories that the caller can ptrace. 2208 2209gid= defines a group authorized to learn processes information otherwise 2210prohibited by hidepid=. If you use some daemon like identd which needs to learn 2211information about processes information, just add identd to this group. 2212 2213subset=pid hides all top level files and directories in the procfs that 2214are not related to tasks. 2215 2216Chapter 5: Filesystem behavior 2217============================== 2218 2219Originally, before the advent of pid namepsace, procfs was a global file 2220system. It means that there was only one procfs instance in the system. 2221 2222When pid namespace was added, a separate procfs instance was mounted in 2223each pid namespace. So, procfs mount options are global among all 2224mountpoints within the same namespace:: 2225 2226 # grep ^proc /proc/mounts 2227 proc /proc proc rw,relatime,hidepid=2 0 0 2228 2229 # strace -e mount mount -o hidepid=1 -t proc proc /tmp/proc 2230 mount("proc", "/tmp/proc", "proc", 0, "hidepid=1") = 0 2231 +++ exited with 0 +++ 2232 2233 # grep ^proc /proc/mounts 2234 proc /proc proc rw,relatime,hidepid=2 0 0 2235 proc /tmp/proc proc rw,relatime,hidepid=2 0 0 2236 2237and only after remounting procfs mount options will change at all 2238mountpoints:: 2239 2240 # mount -o remount,hidepid=1 -t proc proc /tmp/proc 2241 2242 # grep ^proc /proc/mounts 2243 proc /proc proc rw,relatime,hidepid=1 0 0 2244 proc /tmp/proc proc rw,relatime,hidepid=1 0 0 2245 2246This behavior is different from the behavior of other filesystems. 2247 2248The new procfs behavior is more like other filesystems. Each procfs mount 2249creates a new procfs instance. Mount options affect own procfs instance. 2250It means that it became possible to have several procfs instances 2251displaying tasks with different filtering options in one pid namespace:: 2252 2253 # mount -o hidepid=invisible -t proc proc /proc 2254 # mount -o hidepid=noaccess -t proc proc /tmp/proc 2255 # grep ^proc /proc/mounts 2256 proc /proc proc rw,relatime,hidepid=invisible 0 0 2257 proc /tmp/proc proc rw,relatime,hidepid=noaccess 0 0 2258