1------------------------------------------------------------------------------ 2 T H E /proc F I L E S Y S T E M 3------------------------------------------------------------------------------ 4/proc/sys Terrehon Bowden <terrehon@pacbell.net> October 7 1999 5 Bodo Bauer <bb@ricochet.net> 6 72.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000 8move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009 9------------------------------------------------------------------------------ 10Version 1.3 Kernel version 2.2.12 11 Kernel version 2.4.0-test11-pre4 12------------------------------------------------------------------------------ 13fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009 14 15Table of Contents 16----------------- 17 18 0 Preface 19 0.1 Introduction/Credits 20 0.2 Legal Stuff 21 22 1 Collecting System Information 23 1.1 Process-Specific Subdirectories 24 1.2 Kernel data 25 1.3 IDE devices in /proc/ide 26 1.4 Networking info in /proc/net 27 1.5 SCSI info 28 1.6 Parallel port info in /proc/parport 29 1.7 TTY info in /proc/tty 30 1.8 Miscellaneous kernel statistics in /proc/stat 31 1.9 Ext4 file system parameters 32 33 2 Modifying System Parameters 34 35 3 Per-Process Parameters 36 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj - Adjust the oom-killer 37 score 38 3.2 /proc/<pid>/oom_score - Display current oom-killer score 39 3.3 /proc/<pid>/io - Display the IO accounting fields 40 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings 41 3.5 /proc/<pid>/mountinfo - Information about mounts 42 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 43 3.7 /proc/<pid>/task/<tid>/children - Information about task children 44 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file 45 3.9 /proc/<pid>/map_files - Information about memory mapped files 46 3.10 /proc/<pid>/timerslack_ns - Task timerslack value 47 3.11 /proc/<pid>/patch_state - Livepatch patch operation state 48 3.12 /proc/<pid>/arch_status - Task architecture specific information 49 50 4 Configuring procfs 51 4.1 Mount options 52 53------------------------------------------------------------------------------ 54Preface 55------------------------------------------------------------------------------ 56 570.1 Introduction/Credits 58------------------------ 59 60This documentation is part of a soon (or so we hope) to be released book on 61the SuSE Linux distribution. As there is no complete documentation for the 62/proc file system and we've used many freely available sources to write these 63chapters, it seems only fair to give the work back to the Linux community. 64This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm 65afraid it's still far from complete, but we hope it will be useful. As far as 66we know, it is the first 'all-in-one' document about the /proc file system. It 67is focused on the Intel x86 hardware, so if you are looking for PPC, ARM, 68SPARC, AXP, etc., features, you probably won't find what you are looking for. 69It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But 70additions and patches are welcome and will be added to this document if you 71mail them to Bodo. 72 73We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of 74other people for help compiling this documentation. We'd also like to extend a 75special thank you to Andi Kleen for documentation, which we relied on heavily 76to create this document, as well as the additional information he provided. 77Thanks to everybody else who contributed source or docs to the Linux kernel 78and helped create a great piece of software... :) 79 80If you have any comments, corrections or additions, please don't hesitate to 81contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this 82document. 83 84The latest version of this document is available online at 85http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html 86 87If the above direction does not works for you, you could try the kernel 88mailing list at linux-kernel@vger.kernel.org and/or try to reach me at 89comandante@zaralinux.com. 90 910.2 Legal Stuff 92--------------- 93 94We don't guarantee the correctness of this document, and if you come to us 95complaining about how you screwed up your system because of incorrect 96documentation, we won't feel responsible... 97 98------------------------------------------------------------------------------ 99CHAPTER 1: COLLECTING SYSTEM INFORMATION 100------------------------------------------------------------------------------ 101 102------------------------------------------------------------------------------ 103In This Chapter 104------------------------------------------------------------------------------ 105* Investigating the properties of the pseudo file system /proc and its 106 ability to provide information on the running Linux system 107* Examining /proc's structure 108* Uncovering various information about the kernel and the processes running 109 on the system 110------------------------------------------------------------------------------ 111 112 113The proc file system acts as an interface to internal data structures in the 114kernel. It can be used to obtain information about the system and to change 115certain kernel parameters at runtime (sysctl). 116 117First, we'll take a look at the read-only parts of /proc. In Chapter 2, we 118show you how you can use /proc/sys to change settings. 119 1201.1 Process-Specific Subdirectories 121----------------------------------- 122 123The directory /proc contains (among other things) one subdirectory for each 124process running on the system, which is named after the process ID (PID). 125 126The link self points to the process reading the file system. Each process 127subdirectory has the entries listed in Table 1-1. 128 129Note that an open a file descriptor to /proc/<pid> or to any of its 130contained files or subdirectories does not prevent <pid> being reused 131for some other process in the event that <pid> exits. Operations on 132open /proc/<pid> file descriptors corresponding to dead processes 133never act on any new process that the kernel may, through chance, have 134also assigned the process ID <pid>. Instead, operations on these FDs 135usually fail with ESRCH. 136 137Table 1-1: Process specific entries in /proc 138.............................................................................. 139 File Content 140 clear_refs Clears page referenced bits shown in smaps output 141 cmdline Command line arguments 142 cpu Current and last cpu in which it was executed (2.4)(smp) 143 cwd Link to the current working directory 144 environ Values of environment variables 145 exe Link to the executable of this process 146 fd Directory, which contains all file descriptors 147 maps Memory maps to executables and library files (2.4) 148 mem Memory held by this process 149 root Link to the root directory of this process 150 stat Process status 151 statm Process memory status information 152 status Process status in human readable form 153 wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function 154 symbol the task is blocked in - or "0" if not blocked. 155 pagemap Page table 156 stack Report full stack trace, enable via CONFIG_STACKTRACE 157 smaps An extension based on maps, showing the memory consumption of 158 each mapping and flags associated with it 159 smaps_rollup Accumulated smaps stats for all mappings of the process. This 160 can be derived from smaps, but is faster and more convenient 161 numa_maps An extension based on maps, showing the memory locality and 162 binding policy as well as mem usage (in pages) of each mapping. 163.............................................................................. 164 165For example, to get the status information of a process, all you have to do is 166read the file /proc/PID/status: 167 168 >cat /proc/self/status 169 Name: cat 170 State: R (running) 171 Tgid: 5452 172 Pid: 5452 173 PPid: 743 174 TracerPid: 0 (2.4) 175 Uid: 501 501 501 501 176 Gid: 100 100 100 100 177 FDSize: 256 178 Groups: 100 14 16 179 VmPeak: 5004 kB 180 VmSize: 5004 kB 181 VmLck: 0 kB 182 VmHWM: 476 kB 183 VmRSS: 476 kB 184 RssAnon: 352 kB 185 RssFile: 120 kB 186 RssShmem: 4 kB 187 VmData: 156 kB 188 VmStk: 88 kB 189 VmExe: 68 kB 190 VmLib: 1412 kB 191 VmPTE: 20 kb 192 VmSwap: 0 kB 193 HugetlbPages: 0 kB 194 CoreDumping: 0 195 THP_enabled: 1 196 Threads: 1 197 SigQ: 0/28578 198 SigPnd: 0000000000000000 199 ShdPnd: 0000000000000000 200 SigBlk: 0000000000000000 201 SigIgn: 0000000000000000 202 SigCgt: 0000000000000000 203 CapInh: 00000000fffffeff 204 CapPrm: 0000000000000000 205 CapEff: 0000000000000000 206 CapBnd: ffffffffffffffff 207 CapAmb: 0000000000000000 208 NoNewPrivs: 0 209 Seccomp: 0 210 Speculation_Store_Bypass: thread vulnerable 211 voluntary_ctxt_switches: 0 212 nonvoluntary_ctxt_switches: 1 213 214This shows you nearly the same information you would get if you viewed it with 215the ps command. In fact, ps uses the proc file system to obtain its 216information. But you get a more detailed view of the process by reading the 217file /proc/PID/status. It fields are described in table 1-2. 218 219The statm file contains more detailed information about the process 220memory usage. Its seven fields are explained in Table 1-3. The stat file 221contains details information about the process itself. Its fields are 222explained in Table 1-4. 223 224(for SMP CONFIG users) 225For making accounting scalable, RSS related information are handled in an 226asynchronous manner and the value may not be very precise. To see a precise 227snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table. 228It's slow but very precise. 229 230Table 1-2: Contents of the status files (as of 4.19) 231.............................................................................. 232 Field Content 233 Name filename of the executable 234 Umask file mode creation mask 235 State state (R is running, S is sleeping, D is sleeping 236 in an uninterruptible wait, Z is zombie, 237 T is traced or stopped) 238 Tgid thread group ID 239 Ngid NUMA group ID (0 if none) 240 Pid process id 241 PPid process id of the parent process 242 TracerPid PID of process tracing this process (0 if not) 243 Uid Real, effective, saved set, and file system UIDs 244 Gid Real, effective, saved set, and file system GIDs 245 FDSize number of file descriptor slots currently allocated 246 Groups supplementary group list 247 NStgid descendant namespace thread group ID hierarchy 248 NSpid descendant namespace process ID hierarchy 249 NSpgid descendant namespace process group ID hierarchy 250 NSsid descendant namespace session ID hierarchy 251 VmPeak peak virtual memory size 252 VmSize total program size 253 VmLck locked memory size 254 VmPin pinned memory size 255 VmHWM peak resident set size ("high water mark") 256 VmRSS size of memory portions. It contains the three 257 following parts (VmRSS = RssAnon + RssFile + RssShmem) 258 RssAnon size of resident anonymous memory 259 RssFile size of resident file mappings 260 RssShmem size of resident shmem memory (includes SysV shm, 261 mapping of tmpfs and shared anonymous mappings) 262 VmData size of private data segments 263 VmStk size of stack segments 264 VmExe size of text segment 265 VmLib size of shared library code 266 VmPTE size of page table entries 267 VmSwap amount of swap used by anonymous private data 268 (shmem swap usage is not included) 269 HugetlbPages size of hugetlb memory portions 270 CoreDumping process's memory is currently being dumped 271 (killing the process may lead to a corrupted core) 272 THP_enabled process is allowed to use THP (returns 0 when 273 PR_SET_THP_DISABLE is set on the process 274 Threads number of threads 275 SigQ number of signals queued/max. number for queue 276 SigPnd bitmap of pending signals for the thread 277 ShdPnd bitmap of shared pending signals for the process 278 SigBlk bitmap of blocked signals 279 SigIgn bitmap of ignored signals 280 SigCgt bitmap of caught signals 281 CapInh bitmap of inheritable capabilities 282 CapPrm bitmap of permitted capabilities 283 CapEff bitmap of effective capabilities 284 CapBnd bitmap of capabilities bounding set 285 CapAmb bitmap of ambient capabilities 286 NoNewPrivs no_new_privs, like prctl(PR_GET_NO_NEW_PRIV, ...) 287 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...) 288 Speculation_Store_Bypass speculative store bypass mitigation status 289 Cpus_allowed mask of CPUs on which this process may run 290 Cpus_allowed_list Same as previous, but in "list format" 291 Mems_allowed mask of memory nodes allowed to this process 292 Mems_allowed_list Same as previous, but in "list format" 293 voluntary_ctxt_switches number of voluntary context switches 294 nonvoluntary_ctxt_switches number of non voluntary context switches 295.............................................................................. 296 297Table 1-3: Contents of the statm files (as of 2.6.8-rc3) 298.............................................................................. 299 Field Content 300 size total program size (pages) (same as VmSize in status) 301 resident size of memory portions (pages) (same as VmRSS in status) 302 shared number of pages that are shared (i.e. backed by a file, same 303 as RssFile+RssShmem in status) 304 trs number of pages that are 'code' (not including libs; broken, 305 includes data segment) 306 lrs number of pages of library (always 0 on 2.6) 307 drs number of pages of data/stack (including libs; broken, 308 includes library text) 309 dt number of dirty pages (always 0 on 2.6) 310.............................................................................. 311 312 313Table 1-4: Contents of the stat files (as of 2.6.30-rc7) 314.............................................................................. 315 Field Content 316 pid process id 317 tcomm filename of the executable 318 state state (R is running, S is sleeping, D is sleeping in an 319 uninterruptible wait, Z is zombie, T is traced or stopped) 320 ppid process id of the parent process 321 pgrp pgrp of the process 322 sid session id 323 tty_nr tty the process uses 324 tty_pgrp pgrp of the tty 325 flags task flags 326 min_flt number of minor faults 327 cmin_flt number of minor faults with child's 328 maj_flt number of major faults 329 cmaj_flt number of major faults with child's 330 utime user mode jiffies 331 stime kernel mode jiffies 332 cutime user mode jiffies with child's 333 cstime kernel mode jiffies with child's 334 priority priority level 335 nice nice level 336 num_threads number of threads 337 it_real_value (obsolete, always 0) 338 start_time time the process started after system boot 339 vsize virtual memory size 340 rss resident set memory size 341 rsslim current limit in bytes on the rss 342 start_code address above which program text can run 343 end_code address below which program text can run 344 start_stack address of the start of the main process stack 345 esp current value of ESP 346 eip current value of EIP 347 pending bitmap of pending signals 348 blocked bitmap of blocked signals 349 sigign bitmap of ignored signals 350 sigcatch bitmap of caught signals 351 0 (place holder, used to be the wchan address, use /proc/PID/wchan instead) 352 0 (place holder) 353 0 (place holder) 354 exit_signal signal to send to parent thread on exit 355 task_cpu which CPU the task is scheduled on 356 rt_priority realtime priority 357 policy scheduling policy (man sched_setscheduler) 358 blkio_ticks time spent waiting for block IO 359 gtime guest time of the task in jiffies 360 cgtime guest time of the task children in jiffies 361 start_data address above which program data+bss is placed 362 end_data address below which program data+bss is placed 363 start_brk address above which program heap can be expanded with brk() 364 arg_start address above which program command line is placed 365 arg_end address below which program command line is placed 366 env_start address above which program environment is placed 367 env_end address below which program environment is placed 368 exit_code the thread's exit_code in the form reported by the waitpid system call 369.............................................................................. 370 371The /proc/PID/maps file contains the currently mapped memory regions and 372their access permissions. 373 374The format is: 375 376address perms offset dev inode pathname 377 37808048000-08049000 r-xp 00000000 03:00 8312 /opt/test 37908049000-0804a000 rw-p 00001000 03:00 8312 /opt/test 3800804a000-0806b000 rw-p 00000000 00:00 0 [heap] 381a7cb1000-a7cb2000 ---p 00000000 00:00 0 382a7cb2000-a7eb2000 rw-p 00000000 00:00 0 383a7eb2000-a7eb3000 ---p 00000000 00:00 0 384a7eb3000-a7ed5000 rw-p 00000000 00:00 0 385a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6 386a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6 387a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6 388a800b000-a800e000 rw-p 00000000 00:00 0 389a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0 390a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0 391a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0 392a8024000-a8027000 rw-p 00000000 00:00 0 393a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2 394a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2 395a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2 396aff35000-aff4a000 rw-p 00000000 00:00 0 [stack] 397ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso] 398 399where "address" is the address space in the process that it occupies, "perms" 400is a set of permissions: 401 402 r = read 403 w = write 404 x = execute 405 s = shared 406 p = private (copy on write) 407 408"offset" is the offset into the mapping, "dev" is the device (major:minor), and 409"inode" is the inode on that device. 0 indicates that no inode is associated 410with the memory region, as the case would be with BSS (uninitialized data). 411The "pathname" shows the name associated file for this mapping. If the mapping 412is not associated with a file: 413 414 [heap] = the heap of the program 415 [stack] = the stack of the main process 416 [vdso] = the "virtual dynamic shared object", 417 the kernel system call handler 418 [anon:<name>] = an anonymous mapping that has been 419 named by userspace 420 421 or if empty, the mapping is anonymous. 422 423The /proc/PID/smaps is an extension based on maps, showing the memory 424consumption for each of the process's mappings. For each mapping (aka Virtual 425Memory Area, or VMA) there is a series of lines such as the following: 426 42708048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash 428 429Size: 1084 kB 430KernelPageSize: 4 kB 431MMUPageSize: 4 kB 432Rss: 892 kB 433Pss: 374 kB 434Shared_Clean: 892 kB 435Shared_Dirty: 0 kB 436Private_Clean: 0 kB 437Private_Dirty: 0 kB 438Referenced: 892 kB 439Anonymous: 0 kB 440LazyFree: 0 kB 441AnonHugePages: 0 kB 442ShmemPmdMapped: 0 kB 443Shared_Hugetlb: 0 kB 444Private_Hugetlb: 0 kB 445Swap: 0 kB 446SwapPss: 0 kB 447KernelPageSize: 4 kB 448MMUPageSize: 4 kB 449Locked: 0 kB 450THPeligible: 0 451VmFlags: rd ex mr mw me dw 452Name: name from userspace 453 454The first of these lines shows the same information as is displayed for the 455mapping in /proc/PID/maps. Following lines show the size of the mapping 456(size); the size of each page allocated when backing a VMA (KernelPageSize), 457which is usually the same as the size in the page table entries; the page size 458used by the MMU when backing a VMA (in most cases, the same as KernelPageSize); 459the amount of the mapping that is currently resident in RAM (RSS); the 460process' proportional share of this mapping (PSS); and the number of clean and 461dirty shared and private pages in the mapping. 462 463The "proportional set size" (PSS) of a process is the count of pages it has 464in memory, where each page is divided by the number of processes sharing it. 465So if a process has 1000 pages all to itself, and 1000 shared with one other 466process, its PSS will be 1500. 467Note that even a page which is part of a MAP_SHARED mapping, but has only 468a single pte mapped, i.e. is currently used by only one process, is accounted 469as private and not as shared. 470"Referenced" indicates the amount of memory currently marked as referenced or 471accessed. 472"Anonymous" shows the amount of memory that does not belong to any file. Even 473a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE 474and a page is modified, the file page is replaced by a private anonymous copy. 475"LazyFree" shows the amount of memory which is marked by madvise(MADV_FREE). 476The memory isn't freed immediately with madvise(). It's freed in memory 477pressure if the memory is clean. Please note that the printed value might 478be lower than the real value due to optimizations used in the current 479implementation. If this is not desirable please file a bug report. 480"AnonHugePages" shows the ammount of memory backed by transparent hugepage. 481"ShmemPmdMapped" shows the ammount of shared (shmem/tmpfs) memory backed by 482huge pages. 483"Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by 484hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical 485reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field. 486"Swap" shows how much would-be-anonymous memory is also used, but out on swap. 487For shmem mappings, "Swap" includes also the size of the mapped (and not 488replaced by copy-on-write) part of the underlying shmem object out on swap. 489"SwapPss" shows proportional swap share of this mapping. Unlike "Swap", this 490does not take into account swapped out page of underlying shmem objects. 491"Locked" indicates whether the mapping is locked in memory or not. 492"THPeligible" indicates whether the mapping is eligible for allocating THP 493pages - 1 if true, 0 otherwise. It just shows the current status. 494 495"VmFlags" field deserves a separate description. This member represents the kernel 496flags associated with the particular virtual memory area in two letter encoded 497manner. The codes are the following: 498 rd - readable 499 wr - writeable 500 ex - executable 501 sh - shared 502 mr - may read 503 mw - may write 504 me - may execute 505 ms - may share 506 gd - stack segment growns down 507 pf - pure PFN range 508 dw - disabled write to the mapped file 509 lo - pages are locked in memory 510 io - memory mapped I/O area 511 sr - sequential read advise provided 512 rr - random read advise provided 513 dc - do not copy area on fork 514 de - do not expand area on remapping 515 ac - area is accountable 516 nr - swap space is not reserved for the area 517 ht - area uses huge tlb pages 518 ar - architecture specific flag 519 dd - do not include area into core dump 520 sd - soft-dirty flag 521 mm - mixed map area 522 hg - huge page advise flag 523 nh - no-huge page advise flag 524 mg - mergable advise flag 525 526Note that there is no guarantee that every flag and associated mnemonic will 527be present in all further kernel releases. Things get changed, the flags may 528be vanished or the reverse -- new added. Interpretation of their meaning 529might change in future as well. So each consumer of these flags has to 530follow each specific kernel version for the exact semantic. 531 532The "Name" field will only be present on a mapping that has been named by 533userspace, and will show the name passed in by userspace. 534 535This file is only present if the CONFIG_MMU kernel configuration option is 536enabled. 537 538Note: reading /proc/PID/maps or /proc/PID/smaps is inherently racy (consistent 539output can be achieved only in the single read call). 540This typically manifests when doing partial reads of these files while the 541memory map is being modified. Despite the races, we do provide the following 542guarantees: 543 5441) The mapped addresses never go backwards, which implies no two 545 regions will ever overlap. 5462) If there is something at a given vaddr during the entirety of the 547 life of the smaps/maps walk, there will be some output for it. 548 549The /proc/PID/smaps_rollup file includes the same fields as /proc/PID/smaps, 550but their values are the sums of the corresponding values for all mappings of 551the process. Additionally, it contains these fields: 552 553Pss_Anon 554Pss_File 555Pss_Shmem 556 557They represent the proportional shares of anonymous, file, and shmem pages, as 558described for smaps above. These fields are omitted in smaps since each 559mapping identifies the type (anon, file, or shmem) of all pages it contains. 560Thus all information in smaps_rollup can be derived from smaps, but at a 561significantly higher cost. 562 563The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG 564bits on both physical and virtual pages associated with a process, and the 565soft-dirty bit on pte (see Documentation/admin-guide/mm/soft-dirty.rst 566for details). 567To clear the bits for all the pages associated with the process 568 > echo 1 > /proc/PID/clear_refs 569 570To clear the bits for the anonymous pages associated with the process 571 > echo 2 > /proc/PID/clear_refs 572 573To clear the bits for the file mapped pages associated with the process 574 > echo 3 > /proc/PID/clear_refs 575 576To clear the soft-dirty bit 577 > echo 4 > /proc/PID/clear_refs 578 579To reset the peak resident set size ("high water mark") to the process's 580current value: 581 > echo 5 > /proc/PID/clear_refs 582 583Any other value written to /proc/PID/clear_refs will have no effect. 584 585The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags 586using /proc/kpageflags and number of times a page is mapped using 587/proc/kpagecount. For detailed explanation, see 588Documentation/admin-guide/mm/pagemap.rst. 589 590The /proc/pid/numa_maps is an extension based on maps, showing the memory 591locality and binding policy, as well as the memory usage (in pages) of 592each mapping. The output follows a general format where mapping details get 593summarized separated by blank spaces, one mapping per each file line: 594 595address policy mapping details 596 59700400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4 59800600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4 5993206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4 600320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 6013206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 6023206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4 6033206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4 604320698b000 default file=/lib64/libc-2.12.so 6053206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4 6063206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 6073206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4 6087f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4 6097f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4 6107f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048 6117fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4 6127fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4 613 614Where: 615"address" is the starting address for the mapping; 616"policy" reports the NUMA memory policy set for the mapping (see Documentation/admin-guide/mm/numa_memory_policy.rst); 617"mapping details" summarizes mapping data such as mapping type, page usage counters, 618node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page 619size, in KB, that is backing the mapping up. 620 6211.2 Kernel data 622--------------- 623 624Similar to the process entries, the kernel data files give information about 625the running kernel. The files used to obtain this information are contained in 626/proc and are listed in Table 1-5. Not all of these will be present in your 627system. It depends on the kernel configuration and the loaded modules, which 628files are there, and which are missing. 629 630Table 1-5: Kernel info in /proc 631.............................................................................. 632 File Content 633 apm Advanced power management info 634 buddyinfo Kernel memory allocator information (see text) (2.5) 635 bus Directory containing bus specific information 636 cmdline Kernel command line 637 cpuinfo Info about the CPU 638 devices Available devices (block and character) 639 dma Used DMS channels 640 filesystems Supported filesystems 641 driver Various drivers grouped here, currently rtc (2.4) 642 execdomains Execdomains, related to security (2.4) 643 fb Frame Buffer devices (2.4) 644 fs File system parameters, currently nfs/exports (2.4) 645 ide Directory containing info about the IDE subsystem 646 interrupts Interrupt usage 647 iomem Memory map (2.4) 648 ioports I/O port usage 649 irq Masks for irq to cpu affinity (2.4)(smp?) 650 isapnp ISA PnP (Plug&Play) Info (2.4) 651 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4)) 652 kmsg Kernel messages 653 ksyms Kernel symbol table 654 loadavg Load average of last 1, 5 & 15 minutes 655 locks Kernel locks 656 meminfo Memory info 657 misc Miscellaneous 658 modules List of loaded modules 659 mounts Mounted filesystems 660 net Networking info (see text) 661 pagetypeinfo Additional page allocator information (see text) (2.5) 662 partitions Table of partitions known to the system 663 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/, 664 decoupled by lspci (2.4) 665 rtc Real time clock 666 scsi SCSI info (see text) 667 slabinfo Slab pool info 668 softirqs softirq usage 669 stat Overall statistics 670 swaps Swap space utilization 671 sys See chapter 2 672 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4) 673 tty Info of tty drivers 674 uptime Wall clock since boot, combined idle time of all cpus 675 version Kernel version 676 video bttv info of video resources (2.4) 677 vmallocinfo Show vmalloced areas 678.............................................................................. 679 680You can, for example, check which interrupts are currently in use and what 681they are used for by looking in the file /proc/interrupts: 682 683 > cat /proc/interrupts 684 CPU0 685 0: 8728810 XT-PIC timer 686 1: 895 XT-PIC keyboard 687 2: 0 XT-PIC cascade 688 3: 531695 XT-PIC aha152x 689 4: 2014133 XT-PIC serial 690 5: 44401 XT-PIC pcnet_cs 691 8: 2 XT-PIC rtc 692 11: 8 XT-PIC i82365 693 12: 182918 XT-PIC PS/2 Mouse 694 13: 1 XT-PIC fpu 695 14: 1232265 XT-PIC ide0 696 15: 7 XT-PIC ide1 697 NMI: 0 698 699In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the 700output of a SMP machine): 701 702 > cat /proc/interrupts 703 704 CPU0 CPU1 705 0: 1243498 1214548 IO-APIC-edge timer 706 1: 8949 8958 IO-APIC-edge keyboard 707 2: 0 0 XT-PIC cascade 708 5: 11286 10161 IO-APIC-edge soundblaster 709 8: 1 0 IO-APIC-edge rtc 710 9: 27422 27407 IO-APIC-edge 3c503 711 12: 113645 113873 IO-APIC-edge PS/2 Mouse 712 13: 0 0 XT-PIC fpu 713 14: 22491 24012 IO-APIC-edge ide0 714 15: 2183 2415 IO-APIC-edge ide1 715 17: 30564 30414 IO-APIC-level eth0 716 18: 177 164 IO-APIC-level bttv 717 NMI: 2457961 2457959 718 LOC: 2457882 2457881 719 ERR: 2155 720 721NMI is incremented in this case because every timer interrupt generates a NMI 722(Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups. 723 724LOC is the local interrupt counter of the internal APIC of every CPU. 725 726ERR is incremented in the case of errors in the IO-APIC bus (the bus that 727connects the CPUs in a SMP system. This means that an error has been detected, 728the IO-APIC automatically retry the transmission, so it should not be a big 729problem, but you should read the SMP-FAQ. 730 731In 2.6.2* /proc/interrupts was expanded again. This time the goal was for 732/proc/interrupts to display every IRQ vector in use by the system, not 733just those considered 'most important'. The new vectors are: 734 735 THR -- interrupt raised when a machine check threshold counter 736 (typically counting ECC corrected errors of memory or cache) exceeds 737 a configurable threshold. Only available on some systems. 738 739 TRM -- a thermal event interrupt occurs when a temperature threshold 740 has been exceeded for the CPU. This interrupt may also be generated 741 when the temperature drops back to normal. 742 743 SPU -- a spurious interrupt is some interrupt that was raised then lowered 744 by some IO device before it could be fully processed by the APIC. Hence 745 the APIC sees the interrupt but does not know what device it came from. 746 For this case the APIC will generate the interrupt with a IRQ vector 747 of 0xff. This might also be generated by chipset bugs. 748 749 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are 750 sent from one CPU to another per the needs of the OS. Typically, 751 their statistics are used by kernel developers and interested users to 752 determine the occurrence of interrupts of the given type. 753 754The above IRQ vectors are displayed only when relevant. For example, 755the threshold vector does not exist on x86_64 platforms. Others are 756suppressed when the system is a uniprocessor. As of this writing, only 757i386 and x86_64 platforms support the new IRQ vector displays. 758 759Of some interest is the introduction of the /proc/irq directory to 2.4. 760It could be used to set IRQ to CPU affinity, this means that you can "hook" an 761IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the 762irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and 763prof_cpu_mask. 764 765For example 766 > ls /proc/irq/ 767 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask 768 1 11 13 15 17 19 3 5 7 9 default_smp_affinity 769 > ls /proc/irq/0/ 770 smp_affinity 771 772smp_affinity is a bitmask, in which you can specify which CPUs can handle the 773IRQ, you can set it by doing: 774 775 > echo 1 > /proc/irq/10/smp_affinity 776 777This means that only the first CPU will handle the IRQ, but you can also echo 7785 which means that only the first and third CPU can handle the IRQ. 779 780The contents of each smp_affinity file is the same by default: 781 782 > cat /proc/irq/0/smp_affinity 783 ffffffff 784 785There is an alternate interface, smp_affinity_list which allows specifying 786a cpu range instead of a bitmask: 787 788 > cat /proc/irq/0/smp_affinity_list 789 1024-1031 790 791The default_smp_affinity mask applies to all non-active IRQs, which are the 792IRQs which have not yet been allocated/activated, and hence which lack a 793/proc/irq/[0-9]* directory. 794 795The node file on an SMP system shows the node to which the device using the IRQ 796reports itself as being attached. This hardware locality information does not 797include information about any possible driver locality preference. 798 799prof_cpu_mask specifies which CPUs are to be profiled by the system wide 800profiler. Default value is ffffffff (all cpus if there are only 32 of them). 801 802The way IRQs are routed is handled by the IO-APIC, and it's Round Robin 803between all the CPUs which are allowed to handle it. As usual the kernel has 804more info than you and does a better job than you, so the defaults are the 805best choice for almost everyone. [Note this applies only to those IO-APIC's 806that support "Round Robin" interrupt distribution.] 807 808There are three more important subdirectories in /proc: net, scsi, and sys. 809The general rule is that the contents, or even the existence of these 810directories, depend on your kernel configuration. If SCSI is not enabled, the 811directory scsi may not exist. The same is true with the net, which is there 812only when networking support is present in the running kernel. 813 814The slabinfo file gives information about memory usage at the slab level. 815Linux uses slab pools for memory management above page level in version 2.2. 816Commonly used objects have their own slab pool (such as network buffers, 817directory cache, and so on). 818 819.............................................................................. 820 821> cat /proc/buddyinfo 822 823Node 0, zone DMA 0 4 5 4 4 3 ... 824Node 0, zone Normal 1 0 0 1 101 8 ... 825Node 0, zone HighMem 2 0 0 1 1 0 ... 826 827External fragmentation is a problem under some workloads, and buddyinfo is a 828useful tool for helping diagnose these problems. Buddyinfo will give you a 829clue as to how big an area you can safely allocate, or why a previous 830allocation failed. 831 832Each column represents the number of pages of a certain order which are 833available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in 834ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE 835available in ZONE_NORMAL, etc... 836 837More information relevant to external fragmentation can be found in 838pagetypeinfo. 839 840> cat /proc/pagetypeinfo 841Page block order: 9 842Pages per block: 512 843 844Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10 845Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0 846Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0 847Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2 848Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0 849Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0 850Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9 851Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0 852Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452 853Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0 854Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0 855 856Number of blocks type Unmovable Reclaimable Movable Reserve Isolate 857Node 0, zone DMA 2 0 5 1 0 858Node 0, zone DMA32 41 6 967 2 0 859 860Fragmentation avoidance in the kernel works by grouping pages of different 861migrate types into the same contiguous regions of memory called page blocks. 862A page block is typically the size of the default hugepage size e.g. 2MB on 863X86-64. By keeping pages grouped based on their ability to move, the kernel 864can reclaim pages within a page block to satisfy a high-order allocation. 865 866The pagetypinfo begins with information on the size of a page block. It 867then gives the same type of information as buddyinfo except broken down 868by migrate-type and finishes with details on how many page blocks of each 869type exist. 870 871If min_free_kbytes has been tuned correctly (recommendations made by hugeadm 872from libhugetlbfs https://github.com/libhugetlbfs/libhugetlbfs/), one can 873make an estimate of the likely number of huge pages that can be allocated 874at a given point in time. All the "Movable" blocks should be allocatable 875unless memory has been mlock()'d. Some of the Reclaimable blocks should 876also be allocatable although a lot of filesystem metadata may have to be 877reclaimed to achieve this. 878 879.............................................................................. 880 881meminfo: 882 883Provides information about distribution and utilization of memory. This 884varies by architecture and compile options. The following is from a 88516GB PIII, which has highmem enabled. You may not have all of these fields. 886 887> cat /proc/meminfo 888 889MemTotal: 16344972 kB 890MemFree: 13634064 kB 891MemAvailable: 14836172 kB 892Buffers: 3656 kB 893Cached: 1195708 kB 894SwapCached: 0 kB 895Active: 891636 kB 896Inactive: 1077224 kB 897HighTotal: 15597528 kB 898HighFree: 13629632 kB 899LowTotal: 747444 kB 900LowFree: 4432 kB 901SwapTotal: 0 kB 902SwapFree: 0 kB 903Dirty: 968 kB 904Writeback: 0 kB 905AnonPages: 861800 kB 906Mapped: 280372 kB 907Shmem: 644 kB 908KReclaimable: 168048 kB 909Slab: 284364 kB 910SReclaimable: 159856 kB 911SUnreclaim: 124508 kB 912PageTables: 24448 kB 913NFS_Unstable: 0 kB 914Bounce: 0 kB 915WritebackTmp: 0 kB 916CommitLimit: 7669796 kB 917Committed_AS: 100056 kB 918VmallocTotal: 112216 kB 919VmallocUsed: 428 kB 920VmallocChunk: 111088 kB 921Percpu: 62080 kB 922HardwareCorrupted: 0 kB 923AnonHugePages: 49152 kB 924ShmemHugePages: 0 kB 925ShmemPmdMapped: 0 kB 926 927 928 MemTotal: Total usable ram (i.e. physical ram minus a few reserved 929 bits and the kernel binary code) 930 MemFree: The sum of LowFree+HighFree 931MemAvailable: An estimate of how much memory is available for starting new 932 applications, without swapping. Calculated from MemFree, 933 SReclaimable, the size of the file LRU lists, and the low 934 watermarks in each zone. 935 The estimate takes into account that the system needs some 936 page cache to function well, and that not all reclaimable 937 slab will be reclaimable, due to items being in use. The 938 impact of those factors will vary from system to system. 939 Buffers: Relatively temporary storage for raw disk blocks 940 shouldn't get tremendously large (20MB or so) 941 Cached: in-memory cache for files read from the disk (the 942 pagecache). Doesn't include SwapCached 943 SwapCached: Memory that once was swapped out, is swapped back in but 944 still also is in the swapfile (if memory is needed it 945 doesn't need to be swapped out AGAIN because it is already 946 in the swapfile. This saves I/O) 947 Active: Memory that has been used more recently and usually not 948 reclaimed unless absolutely necessary. 949 Inactive: Memory which has been less recently used. It is more 950 eligible to be reclaimed for other purposes 951 HighTotal: 952 HighFree: Highmem is all memory above ~860MB of physical memory 953 Highmem areas are for use by userspace programs, or 954 for the pagecache. The kernel must use tricks to access 955 this memory, making it slower to access than lowmem. 956 LowTotal: 957 LowFree: Lowmem is memory which can be used for everything that 958 highmem can be used for, but it is also available for the 959 kernel's use for its own data structures. Among many 960 other things, it is where everything from the Slab is 961 allocated. Bad things happen when you're out of lowmem. 962 SwapTotal: total amount of swap space available 963 SwapFree: Memory which has been evicted from RAM, and is temporarily 964 on the disk 965 Dirty: Memory which is waiting to get written back to the disk 966 Writeback: Memory which is actively being written back to the disk 967 AnonPages: Non-file backed pages mapped into userspace page tables 968HardwareCorrupted: The amount of RAM/memory in KB, the kernel identifies as 969 corrupted. 970AnonHugePages: Non-file backed huge pages mapped into userspace page tables 971 Mapped: files which have been mmaped, such as libraries 972 Shmem: Total memory used by shared memory (shmem) and tmpfs 973ShmemHugePages: Memory used by shared memory (shmem) and tmpfs allocated 974 with huge pages 975ShmemPmdMapped: Shared memory mapped into userspace with huge pages 976KReclaimable: Kernel allocations that the kernel will attempt to reclaim 977 under memory pressure. Includes SReclaimable (below), and other 978 direct allocations with a shrinker. 979 Slab: in-kernel data structures cache 980SReclaimable: Part of Slab, that might be reclaimed, such as caches 981 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure 982 PageTables: amount of memory dedicated to the lowest level of page 983 tables. 984NFS_Unstable: NFS pages sent to the server, but not yet committed to stable 985 storage 986 Bounce: Memory used for block device "bounce buffers" 987WritebackTmp: Memory used by FUSE for temporary writeback buffers 988 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'), 989 this is the total amount of memory currently available to 990 be allocated on the system. This limit is only adhered to 991 if strict overcommit accounting is enabled (mode 2 in 992 'vm.overcommit_memory'). 993 The CommitLimit is calculated with the following formula: 994 CommitLimit = ([total RAM pages] - [total huge TLB pages]) * 995 overcommit_ratio / 100 + [total swap pages] 996 For example, on a system with 1G of physical RAM and 7G 997 of swap with a `vm.overcommit_ratio` of 30 it would 998 yield a CommitLimit of 7.3G. 999 For more details, see the memory overcommit documentation 1000 in vm/overcommit-accounting. 1001Committed_AS: The amount of memory presently allocated on the system. 1002 The committed memory is a sum of all of the memory which 1003 has been allocated by processes, even if it has not been 1004 "used" by them as of yet. A process which malloc()'s 1G 1005 of memory, but only touches 300M of it will show up as 1006 using 1G. This 1G is memory which has been "committed" to 1007 by the VM and can be used at any time by the allocating 1008 application. With strict overcommit enabled on the system 1009 (mode 2 in 'vm.overcommit_memory'),allocations which would 1010 exceed the CommitLimit (detailed above) will not be permitted. 1011 This is useful if one needs to guarantee that processes will 1012 not fail due to lack of memory once that memory has been 1013 successfully allocated. 1014VmallocTotal: total size of vmalloc memory area 1015 VmallocUsed: amount of vmalloc area which is used 1016VmallocChunk: largest contiguous block of vmalloc area which is free 1017 Percpu: Memory allocated to the percpu allocator used to back percpu 1018 allocations. This stat excludes the cost of metadata. 1019 1020.............................................................................. 1021 1022vmallocinfo: 1023 1024Provides information about vmalloced/vmaped areas. One line per area, 1025containing the virtual address range of the area, size in bytes, 1026caller information of the creator, and optional information depending 1027on the kind of area : 1028 1029 pages=nr number of pages 1030 phys=addr if a physical address was specified 1031 ioremap I/O mapping (ioremap() and friends) 1032 vmalloc vmalloc() area 1033 vmap vmap()ed pages 1034 user VM_USERMAP area 1035 vpages buffer for pages pointers was vmalloced (huge area) 1036 N<node>=nr (Only on NUMA kernels) 1037 Number of pages allocated on memory node <node> 1038 1039> cat /proc/vmallocinfo 10400xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ... 1041 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128 10420xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ... 1043 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64 10440xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f... 1045 phys=7fee8000 ioremap 10460xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f... 1047 phys=7fee7000 ioremap 10480xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210 10490xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ... 1050 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3 10510xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ... 1052 pages=2 vmalloc N1=2 10530xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ... 1054 /0x130 [x_tables] pages=4 vmalloc N0=4 10550xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ... 1056 pages=14 vmalloc N2=14 10570xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ... 1058 pages=4 vmalloc N1=4 10590xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ... 1060 pages=2 vmalloc N1=2 10610xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ... 1062 pages=10 vmalloc N0=10 1063 1064.............................................................................. 1065 1066softirqs: 1067 1068Provides counts of softirq handlers serviced since boot time, for each cpu. 1069 1070> cat /proc/softirqs 1071 CPU0 CPU1 CPU2 CPU3 1072 HI: 0 0 0 0 1073 TIMER: 27166 27120 27097 27034 1074 NET_TX: 0 0 0 17 1075 NET_RX: 42 0 0 39 1076 BLOCK: 0 0 107 1121 1077 TASKLET: 0 0 0 290 1078 SCHED: 27035 26983 26971 26746 1079 HRTIMER: 0 0 0 0 1080 RCU: 1678 1769 2178 2250 1081 1082 10831.3 IDE devices in /proc/ide 1084---------------------------- 1085 1086The subdirectory /proc/ide contains information about all IDE devices of which 1087the kernel is aware. There is one subdirectory for each IDE controller, the 1088file drivers and a link for each IDE device, pointing to the device directory 1089in the controller specific subtree. 1090 1091The file drivers contains general information about the drivers used for the 1092IDE devices: 1093 1094 > cat /proc/ide/drivers 1095 ide-cdrom version 4.53 1096 ide-disk version 1.08 1097 1098More detailed information can be found in the controller specific 1099subdirectories. These are named ide0, ide1 and so on. Each of these 1100directories contains the files shown in table 1-6. 1101 1102 1103Table 1-6: IDE controller info in /proc/ide/ide? 1104.............................................................................. 1105 File Content 1106 channel IDE channel (0 or 1) 1107 config Configuration (only for PCI/IDE bridge) 1108 mate Mate name 1109 model Type/Chipset of IDE controller 1110.............................................................................. 1111 1112Each device connected to a controller has a separate subdirectory in the 1113controllers directory. The files listed in table 1-7 are contained in these 1114directories. 1115 1116 1117Table 1-7: IDE device information 1118.............................................................................. 1119 File Content 1120 cache The cache 1121 capacity Capacity of the medium (in 512Byte blocks) 1122 driver driver and version 1123 geometry physical and logical geometry 1124 identify device identify block 1125 media media type 1126 model device identifier 1127 settings device setup 1128 smart_thresholds IDE disk management thresholds 1129 smart_values IDE disk management values 1130.............................................................................. 1131 1132The most interesting file is settings. This file contains a nice overview of 1133the drive parameters: 1134 1135 # cat /proc/ide/ide0/hda/settings 1136 name value min max mode 1137 ---- ----- --- --- ---- 1138 bios_cyl 526 0 65535 rw 1139 bios_head 255 0 255 rw 1140 bios_sect 63 0 63 rw 1141 breada_readahead 4 0 127 rw 1142 bswap 0 0 1 r 1143 file_readahead 72 0 2097151 rw 1144 io_32bit 0 0 3 rw 1145 keepsettings 0 0 1 rw 1146 max_kb_per_request 122 1 127 rw 1147 multcount 0 0 8 rw 1148 nice1 1 0 1 rw 1149 nowerr 0 0 1 rw 1150 pio_mode write-only 0 255 w 1151 slow 0 0 1 rw 1152 unmaskirq 0 0 1 rw 1153 using_dma 0 0 1 rw 1154 1155 11561.4 Networking info in /proc/net 1157-------------------------------- 1158 1159The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the 1160additional values you get for IP version 6 if you configure the kernel to 1161support this. Table 1-9 lists the files and their meaning. 1162 1163 1164Table 1-8: IPv6 info in /proc/net 1165.............................................................................. 1166 File Content 1167 udp6 UDP sockets (IPv6) 1168 tcp6 TCP sockets (IPv6) 1169 raw6 Raw device statistics (IPv6) 1170 igmp6 IP multicast addresses, which this host joined (IPv6) 1171 if_inet6 List of IPv6 interface addresses 1172 ipv6_route Kernel routing table for IPv6 1173 rt6_stats Global IPv6 routing tables statistics 1174 sockstat6 Socket statistics (IPv6) 1175 snmp6 Snmp data (IPv6) 1176.............................................................................. 1177 1178 1179Table 1-9: Network info in /proc/net 1180.............................................................................. 1181 File Content 1182 arp Kernel ARP table 1183 dev network devices with statistics 1184 dev_mcast the Layer2 multicast groups a device is listening too 1185 (interface index, label, number of references, number of bound 1186 addresses). 1187 dev_stat network device status 1188 ip_fwchains Firewall chain linkage 1189 ip_fwnames Firewall chain names 1190 ip_masq Directory containing the masquerading tables 1191 ip_masquerade Major masquerading table 1192 netstat Network statistics 1193 raw raw device statistics 1194 route Kernel routing table 1195 rpc Directory containing rpc info 1196 rt_cache Routing cache 1197 snmp SNMP data 1198 sockstat Socket statistics 1199 tcp TCP sockets 1200 udp UDP sockets 1201 unix UNIX domain sockets 1202 wireless Wireless interface data (Wavelan etc) 1203 igmp IP multicast addresses, which this host joined 1204 psched Global packet scheduler parameters. 1205 netlink List of PF_NETLINK sockets 1206 ip_mr_vifs List of multicast virtual interfaces 1207 ip_mr_cache List of multicast routing cache 1208.............................................................................. 1209 1210You can use this information to see which network devices are available in 1211your system and how much traffic was routed over those devices: 1212 1213 > cat /proc/net/dev 1214 Inter-|Receive |[... 1215 face |bytes packets errs drop fifo frame compressed multicast|[... 1216 lo: 908188 5596 0 0 0 0 0 0 [... 1217 ppp0:15475140 20721 410 0 0 410 0 0 [... 1218 eth0: 614530 7085 0 0 0 0 0 1 [... 1219 1220 ...] Transmit 1221 ...] bytes packets errs drop fifo colls carrier compressed 1222 ...] 908188 5596 0 0 0 0 0 0 1223 ...] 1375103 17405 0 0 0 0 0 0 1224 ...] 1703981 5535 0 0 0 3 0 0 1225 1226In addition, each Channel Bond interface has its own directory. For 1227example, the bond0 device will have a directory called /proc/net/bond0/. 1228It will contain information that is specific to that bond, such as the 1229current slaves of the bond, the link status of the slaves, and how 1230many times the slaves link has failed. 1231 12321.5 SCSI info 1233------------- 1234 1235If you have a SCSI host adapter in your system, you'll find a subdirectory 1236named after the driver for this adapter in /proc/scsi. You'll also see a list 1237of all recognized SCSI devices in /proc/scsi: 1238 1239 >cat /proc/scsi/scsi 1240 Attached devices: 1241 Host: scsi0 Channel: 00 Id: 00 Lun: 00 1242 Vendor: IBM Model: DGHS09U Rev: 03E0 1243 Type: Direct-Access ANSI SCSI revision: 03 1244 Host: scsi0 Channel: 00 Id: 06 Lun: 00 1245 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04 1246 Type: CD-ROM ANSI SCSI revision: 02 1247 1248 1249The directory named after the driver has one file for each adapter found in 1250the system. These files contain information about the controller, including 1251the used IRQ and the IO address range. The amount of information shown is 1252dependent on the adapter you use. The example shows the output for an Adaptec 1253AHA-2940 SCSI adapter: 1254 1255 > cat /proc/scsi/aic7xxx/0 1256 1257 Adaptec AIC7xxx driver version: 5.1.19/3.2.4 1258 Compile Options: 1259 TCQ Enabled By Default : Disabled 1260 AIC7XXX_PROC_STATS : Disabled 1261 AIC7XXX_RESET_DELAY : 5 1262 Adapter Configuration: 1263 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter 1264 Ultra Wide Controller 1265 PCI MMAPed I/O Base: 0xeb001000 1266 Adapter SEEPROM Config: SEEPROM found and used. 1267 Adaptec SCSI BIOS: Enabled 1268 IRQ: 10 1269 SCBs: Active 0, Max Active 2, 1270 Allocated 15, HW 16, Page 255 1271 Interrupts: 160328 1272 BIOS Control Word: 0x18b6 1273 Adapter Control Word: 0x005b 1274 Extended Translation: Enabled 1275 Disconnect Enable Flags: 0xffff 1276 Ultra Enable Flags: 0x0001 1277 Tag Queue Enable Flags: 0x0000 1278 Ordered Queue Tag Flags: 0x0000 1279 Default Tag Queue Depth: 8 1280 Tagged Queue By Device array for aic7xxx host instance 0: 1281 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255} 1282 Actual queue depth per device for aic7xxx host instance 0: 1283 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} 1284 Statistics: 1285 (scsi0:0:0:0) 1286 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8 1287 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0) 1288 Total transfers 160151 (74577 reads and 85574 writes) 1289 (scsi0:0:6:0) 1290 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15 1291 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0) 1292 Total transfers 0 (0 reads and 0 writes) 1293 1294 12951.6 Parallel port info in /proc/parport 1296--------------------------------------- 1297 1298The directory /proc/parport contains information about the parallel ports of 1299your system. It has one subdirectory for each port, named after the port 1300number (0,1,2,...). 1301 1302These directories contain the four files shown in Table 1-10. 1303 1304 1305Table 1-10: Files in /proc/parport 1306.............................................................................. 1307 File Content 1308 autoprobe Any IEEE-1284 device ID information that has been acquired. 1309 devices list of the device drivers using that port. A + will appear by the 1310 name of the device currently using the port (it might not appear 1311 against any). 1312 hardware Parallel port's base address, IRQ line and DMA channel. 1313 irq IRQ that parport is using for that port. This is in a separate 1314 file to allow you to alter it by writing a new value in (IRQ 1315 number or none). 1316.............................................................................. 1317 13181.7 TTY info in /proc/tty 1319------------------------- 1320 1321Information about the available and actually used tty's can be found in the 1322directory /proc/tty.You'll find entries for drivers and line disciplines in 1323this directory, as shown in Table 1-11. 1324 1325 1326Table 1-11: Files in /proc/tty 1327.............................................................................. 1328 File Content 1329 drivers list of drivers and their usage 1330 ldiscs registered line disciplines 1331 driver/serial usage statistic and status of single tty lines 1332.............................................................................. 1333 1334To see which tty's are currently in use, you can simply look into the file 1335/proc/tty/drivers: 1336 1337 > cat /proc/tty/drivers 1338 pty_slave /dev/pts 136 0-255 pty:slave 1339 pty_master /dev/ptm 128 0-255 pty:master 1340 pty_slave /dev/ttyp 3 0-255 pty:slave 1341 pty_master /dev/pty 2 0-255 pty:master 1342 serial /dev/cua 5 64-67 serial:callout 1343 serial /dev/ttyS 4 64-67 serial 1344 /dev/tty0 /dev/tty0 4 0 system:vtmaster 1345 /dev/ptmx /dev/ptmx 5 2 system 1346 /dev/console /dev/console 5 1 system:console 1347 /dev/tty /dev/tty 5 0 system:/dev/tty 1348 unknown /dev/tty 4 1-63 console 1349 1350 13511.8 Miscellaneous kernel statistics in /proc/stat 1352------------------------------------------------- 1353 1354Various pieces of information about kernel activity are available in the 1355/proc/stat file. All of the numbers reported in this file are aggregates 1356since the system first booted. For a quick look, simply cat the file: 1357 1358 > cat /proc/stat 1359 cpu 2255 34 2290 22625563 6290 127 456 0 0 0 1360 cpu0 1132 34 1441 11311718 3675 127 438 0 0 0 1361 cpu1 1123 0 849 11313845 2614 0 18 0 0 0 1362 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...] 1363 ctxt 1990473 1364 btime 1062191376 1365 processes 2915 1366 procs_running 1 1367 procs_blocked 0 1368 softirq 183433 0 21755 12 39 1137 231 21459 2263 1369 1370The very first "cpu" line aggregates the numbers in all of the other "cpuN" 1371lines. These numbers identify the amount of time the CPU has spent performing 1372different kinds of work. Time units are in USER_HZ (typically hundredths of a 1373second). The meanings of the columns are as follows, from left to right: 1374 1375- user: normal processes executing in user mode 1376- nice: niced processes executing in user mode 1377- system: processes executing in kernel mode 1378- idle: twiddling thumbs 1379- iowait: In a word, iowait stands for waiting for I/O to complete. But there 1380 are several problems: 1381 1. Cpu will not wait for I/O to complete, iowait is the time that a task is 1382 waiting for I/O to complete. When cpu goes into idle state for 1383 outstanding task io, another task will be scheduled on this CPU. 1384 2. In a multi-core CPU, the task waiting for I/O to complete is not running 1385 on any CPU, so the iowait of each CPU is difficult to calculate. 1386 3. The value of iowait field in /proc/stat will decrease in certain 1387 conditions. 1388 So, the iowait is not reliable by reading from /proc/stat. 1389- irq: servicing interrupts 1390- softirq: servicing softirqs 1391- steal: involuntary wait 1392- guest: running a normal guest 1393- guest_nice: running a niced guest 1394 1395The "intr" line gives counts of interrupts serviced since boot time, for each 1396of the possible system interrupts. The first column is the total of all 1397interrupts serviced including unnumbered architecture specific interrupts; 1398each subsequent column is the total for that particular numbered interrupt. 1399Unnumbered interrupts are not shown, only summed into the total. 1400 1401The "ctxt" line gives the total number of context switches across all CPUs. 1402 1403The "btime" line gives the time at which the system booted, in seconds since 1404the Unix epoch. 1405 1406The "processes" line gives the number of processes and threads created, which 1407includes (but is not limited to) those created by calls to the fork() and 1408clone() system calls. 1409 1410The "procs_running" line gives the total number of threads that are 1411running or ready to run (i.e., the total number of runnable threads). 1412 1413The "procs_blocked" line gives the number of processes currently blocked, 1414waiting for I/O to complete. 1415 1416The "softirq" line gives counts of softirqs serviced since boot time, for each 1417of the possible system softirqs. The first column is the total of all 1418softirqs serviced; each subsequent column is the total for that particular 1419softirq. 1420 1421 14221.9 Ext4 file system parameters 1423------------------------------- 1424 1425Information about mounted ext4 file systems can be found in 1426/proc/fs/ext4. Each mounted filesystem will have a directory in 1427/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or 1428/proc/fs/ext4/dm-0). The files in each per-device directory are shown 1429in Table 1-12, below. 1430 1431Table 1-12: Files in /proc/fs/ext4/<devname> 1432.............................................................................. 1433 File Content 1434 mb_groups details of multiblock allocator buddy cache of free blocks 1435.............................................................................. 1436 14372.0 /proc/consoles 1438------------------ 1439Shows registered system console lines. 1440 1441To see which character device lines are currently used for the system console 1442/dev/console, you may simply look into the file /proc/consoles: 1443 1444 > cat /proc/consoles 1445 tty0 -WU (ECp) 4:7 1446 ttyS0 -W- (Ep) 4:64 1447 1448The columns are: 1449 1450 device name of the device 1451 operations R = can do read operations 1452 W = can do write operations 1453 U = can do unblank 1454 flags E = it is enabled 1455 C = it is preferred console 1456 B = it is primary boot console 1457 p = it is used for printk buffer 1458 b = it is not a TTY but a Braille device 1459 a = it is safe to use when cpu is offline 1460 major:minor major and minor number of the device separated by a colon 1461 1462------------------------------------------------------------------------------ 1463Summary 1464------------------------------------------------------------------------------ 1465The /proc file system serves information about the running system. It not only 1466allows access to process data but also allows you to request the kernel status 1467by reading files in the hierarchy. 1468 1469The directory structure of /proc reflects the types of information and makes 1470it easy, if not obvious, where to look for specific data. 1471------------------------------------------------------------------------------ 1472 1473------------------------------------------------------------------------------ 1474CHAPTER 2: MODIFYING SYSTEM PARAMETERS 1475------------------------------------------------------------------------------ 1476 1477------------------------------------------------------------------------------ 1478In This Chapter 1479------------------------------------------------------------------------------ 1480* Modifying kernel parameters by writing into files found in /proc/sys 1481* Exploring the files which modify certain parameters 1482* Review of the /proc/sys file tree 1483------------------------------------------------------------------------------ 1484 1485 1486A very interesting part of /proc is the directory /proc/sys. This is not only 1487a source of information, it also allows you to change parameters within the 1488kernel. Be very careful when attempting this. You can optimize your system, 1489but you can also cause it to crash. Never alter kernel parameters on a 1490production system. Set up a development machine and test to make sure that 1491everything works the way you want it to. You may have no alternative but to 1492reboot the machine once an error has been made. 1493 1494To change a value, simply echo the new value into the file. An example is 1495given below in the section on the file system data. You need to be root to do 1496this. You can create your own boot script to perform this every time your 1497system boots. 1498 1499The files in /proc/sys can be used to fine tune and monitor miscellaneous and 1500general things in the operation of the Linux kernel. Since some of the files 1501can inadvertently disrupt your system, it is advisable to read both 1502documentation and source before actually making adjustments. In any case, be 1503very careful when writing to any of these files. The entries in /proc may 1504change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt 1505review the kernel documentation in the directory /usr/src/linux/Documentation. 1506This chapter is heavily based on the documentation included in the pre 2.2 1507kernels, and became part of it in version 2.2.1 of the Linux kernel. 1508 1509Please see: Documentation/admin-guide/sysctl/ directory for descriptions of these 1510entries. 1511 1512------------------------------------------------------------------------------ 1513Summary 1514------------------------------------------------------------------------------ 1515Certain aspects of kernel behavior can be modified at runtime, without the 1516need to recompile the kernel, or even to reboot the system. The files in the 1517/proc/sys tree can not only be read, but also modified. You can use the echo 1518command to write value into these files, thereby changing the default settings 1519of the kernel. 1520------------------------------------------------------------------------------ 1521 1522------------------------------------------------------------------------------ 1523CHAPTER 3: PER-PROCESS PARAMETERS 1524------------------------------------------------------------------------------ 1525 15263.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score 1527-------------------------------------------------------------------------------- 1528 1529These file can be used to adjust the badness heuristic used to select which 1530process gets killed in out of memory conditions. 1531 1532The badness heuristic assigns a value to each candidate task ranging from 0 1533(never kill) to 1000 (always kill) to determine which process is targeted. The 1534units are roughly a proportion along that range of allowed memory the process 1535may allocate from based on an estimation of its current memory and swap use. 1536For example, if a task is using all allowed memory, its badness score will be 15371000. If it is using half of its allowed memory, its score will be 500. 1538 1539There is an additional factor included in the badness score: the current memory 1540and swap usage is discounted by 3% for root processes. 1541 1542The amount of "allowed" memory depends on the context in which the oom killer 1543was called. If it is due to the memory assigned to the allocating task's cpuset 1544being exhausted, the allowed memory represents the set of mems assigned to that 1545cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed 1546memory represents the set of mempolicy nodes. If it is due to a memory 1547limit (or swap limit) being reached, the allowed memory is that configured 1548limit. Finally, if it is due to the entire system being out of memory, the 1549allowed memory represents all allocatable resources. 1550 1551The value of /proc/<pid>/oom_score_adj is added to the badness score before it 1552is used to determine which task to kill. Acceptable values range from -1000 1553(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to 1554polarize the preference for oom killing either by always preferring a certain 1555task or completely disabling it. The lowest possible value, -1000, is 1556equivalent to disabling oom killing entirely for that task since it will always 1557report a badness score of 0. 1558 1559Consequently, it is very simple for userspace to define the amount of memory to 1560consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for 1561example, is roughly equivalent to allowing the remainder of tasks sharing the 1562same system, cpuset, mempolicy, or memory controller resources to use at least 156350% more memory. A value of -500, on the other hand, would be roughly 1564equivalent to discounting 50% of the task's allowed memory from being considered 1565as scoring against the task. 1566 1567For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also 1568be used to tune the badness score. Its acceptable values range from -16 1569(OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17 1570(OOM_DISABLE) to disable oom killing entirely for that task. Its value is 1571scaled linearly with /proc/<pid>/oom_score_adj. 1572 1573The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last 1574value set by a CAP_SYS_RESOURCE process. To reduce the value any lower 1575requires CAP_SYS_RESOURCE. 1576 1577Caveat: when a parent task is selected, the oom killer will sacrifice any first 1578generation children with separate address spaces instead, if possible. This 1579avoids servers and important system daemons from being killed and loses the 1580minimal amount of work. 1581 1582 15833.2 /proc/<pid>/oom_score - Display current oom-killer score 1584------------------------------------------------------------- 1585 1586This file can be used to check the current score used by the oom-killer is for 1587any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which 1588process should be killed in an out-of-memory situation. 1589 1590 15913.3 /proc/<pid>/io - Display the IO accounting fields 1592------------------------------------------------------- 1593 1594This file contains IO statistics for each running process 1595 1596Example 1597------- 1598 1599test:/tmp # dd if=/dev/zero of=/tmp/test.dat & 1600[1] 3828 1601 1602test:/tmp # cat /proc/3828/io 1603rchar: 323934931 1604wchar: 323929600 1605syscr: 632687 1606syscw: 632675 1607read_bytes: 0 1608write_bytes: 323932160 1609cancelled_write_bytes: 0 1610 1611 1612Description 1613----------- 1614 1615rchar 1616----- 1617 1618I/O counter: chars read 1619The number of bytes which this task has caused to be read from storage. This 1620is simply the sum of bytes which this process passed to read() and pread(). 1621It includes things like tty IO and it is unaffected by whether or not actual 1622physical disk IO was required (the read might have been satisfied from 1623pagecache) 1624 1625 1626wchar 1627----- 1628 1629I/O counter: chars written 1630The number of bytes which this task has caused, or shall cause to be written 1631to disk. Similar caveats apply here as with rchar. 1632 1633 1634syscr 1635----- 1636 1637I/O counter: read syscalls 1638Attempt to count the number of read I/O operations, i.e. syscalls like read() 1639and pread(). 1640 1641 1642syscw 1643----- 1644 1645I/O counter: write syscalls 1646Attempt to count the number of write I/O operations, i.e. syscalls like 1647write() and pwrite(). 1648 1649 1650read_bytes 1651---------- 1652 1653I/O counter: bytes read 1654Attempt to count the number of bytes which this process really did cause to 1655be fetched from the storage layer. Done at the submit_bio() level, so it is 1656accurate for block-backed filesystems. <please add status regarding NFS and 1657CIFS at a later time> 1658 1659 1660write_bytes 1661----------- 1662 1663I/O counter: bytes written 1664Attempt to count the number of bytes which this process caused to be sent to 1665the storage layer. This is done at page-dirtying time. 1666 1667 1668cancelled_write_bytes 1669--------------------- 1670 1671The big inaccuracy here is truncate. If a process writes 1MB to a file and 1672then deletes the file, it will in fact perform no writeout. But it will have 1673been accounted as having caused 1MB of write. 1674In other words: The number of bytes which this process caused to not happen, 1675by truncating pagecache. A task can cause "negative" IO too. If this task 1676truncates some dirty pagecache, some IO which another task has been accounted 1677for (in its write_bytes) will not be happening. We _could_ just subtract that 1678from the truncating task's write_bytes, but there is information loss in doing 1679that. 1680 1681 1682Note 1683---- 1684 1685At its current implementation state, this is a bit racy on 32-bit machines: if 1686process A reads process B's /proc/pid/io while process B is updating one of 1687those 64-bit counters, process A could see an intermediate result. 1688 1689 1690More information about this can be found within the taskstats documentation in 1691Documentation/accounting. 1692 16933.4 /proc/<pid>/coredump_filter - Core dump filtering settings 1694--------------------------------------------------------------- 1695When a process is dumped, all anonymous memory is written to a core file as 1696long as the size of the core file isn't limited. But sometimes we don't want 1697to dump some memory segments, for example, huge shared memory or DAX. 1698Conversely, sometimes we want to save file-backed memory segments into a core 1699file, not only the individual files. 1700 1701/proc/<pid>/coredump_filter allows you to customize which memory segments 1702will be dumped when the <pid> process is dumped. coredump_filter is a bitmask 1703of memory types. If a bit of the bitmask is set, memory segments of the 1704corresponding memory type are dumped, otherwise they are not dumped. 1705 1706The following 9 memory types are supported: 1707 - (bit 0) anonymous private memory 1708 - (bit 1) anonymous shared memory 1709 - (bit 2) file-backed private memory 1710 - (bit 3) file-backed shared memory 1711 - (bit 4) ELF header pages in file-backed private memory areas (it is 1712 effective only if the bit 2 is cleared) 1713 - (bit 5) hugetlb private memory 1714 - (bit 6) hugetlb shared memory 1715 - (bit 7) DAX private memory 1716 - (bit 8) DAX shared memory 1717 1718 Note that MMIO pages such as frame buffer are never dumped and vDSO pages 1719 are always dumped regardless of the bitmask status. 1720 1721 Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is 1722 only affected by bit 5-6, and DAX is only affected by bits 7-8. 1723 1724The default value of coredump_filter is 0x33; this means all anonymous memory 1725segments, ELF header pages and hugetlb private memory are dumped. 1726 1727If you don't want to dump all shared memory segments attached to pid 1234, 1728write 0x31 to the process's proc file. 1729 1730 $ echo 0x31 > /proc/1234/coredump_filter 1731 1732When a new process is created, the process inherits the bitmask status from its 1733parent. It is useful to set up coredump_filter before the program runs. 1734For example: 1735 1736 $ echo 0x7 > /proc/self/coredump_filter 1737 $ ./some_program 1738 17393.5 /proc/<pid>/mountinfo - Information about mounts 1740-------------------------------------------------------- 1741 1742This file contains lines of the form: 1743 174436 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue 1745(1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11) 1746 1747(1) mount ID: unique identifier of the mount (may be reused after umount) 1748(2) parent ID: ID of parent (or of self for the top of the mount tree) 1749(3) major:minor: value of st_dev for files on filesystem 1750(4) root: root of the mount within the filesystem 1751(5) mount point: mount point relative to the process's root 1752(6) mount options: per mount options 1753(7) optional fields: zero or more fields of the form "tag[:value]" 1754(8) separator: marks the end of the optional fields 1755(9) filesystem type: name of filesystem of the form "type[.subtype]" 1756(10) mount source: filesystem specific information or "none" 1757(11) super options: per super block options 1758 1759Parsers should ignore all unrecognised optional fields. Currently the 1760possible optional fields are: 1761 1762shared:X mount is shared in peer group X 1763master:X mount is slave to peer group X 1764propagate_from:X mount is slave and receives propagation from peer group X (*) 1765unbindable mount is unbindable 1766 1767(*) X is the closest dominant peer group under the process's root. If 1768X is the immediate master of the mount, or if there's no dominant peer 1769group under the same root, then only the "master:X" field is present 1770and not the "propagate_from:X" field. 1771 1772For more information on mount propagation see: 1773 1774 Documentation/filesystems/sharedsubtree.txt 1775 1776 17773.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 1778-------------------------------------------------------- 1779These files provide a method to access a tasks comm value. It also allows for 1780a task to set its own or one of its thread siblings comm value. The comm value 1781is limited in size compared to the cmdline value, so writing anything longer 1782then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated 1783comm value. 1784 1785 17863.7 /proc/<pid>/task/<tid>/children - Information about task children 1787------------------------------------------------------------------------- 1788This file provides a fast way to retrieve first level children pids 1789of a task pointed by <pid>/<tid> pair. The format is a space separated 1790stream of pids. 1791 1792Note the "first level" here -- if a child has own children they will 1793not be listed here, one needs to read /proc/<children-pid>/task/<tid>/children 1794to obtain the descendants. 1795 1796Since this interface is intended to be fast and cheap it doesn't 1797guarantee to provide precise results and some children might be 1798skipped, especially if they've exited right after we printed their 1799pids, so one need to either stop or freeze processes being inspected 1800if precise results are needed. 1801 1802 18033.8 /proc/<pid>/fdinfo/<fd> - Information about opened file 1804--------------------------------------------------------------- 1805This file provides information associated with an opened file. The regular 1806files have at least three fields -- 'pos', 'flags' and mnt_id. The 'pos' 1807represents the current offset of the opened file in decimal form [see lseek(2) 1808for details], 'flags' denotes the octal O_xxx mask the file has been 1809created with [see open(2) for details] and 'mnt_id' represents mount ID of 1810the file system containing the opened file [see 3.5 /proc/<pid>/mountinfo 1811for details]. 1812 1813A typical output is 1814 1815 pos: 0 1816 flags: 0100002 1817 mnt_id: 19 1818 1819All locks associated with a file descriptor are shown in its fdinfo too. 1820 1821lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF 1822 1823The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags 1824pair provide additional information particular to the objects they represent. 1825 1826 Eventfd files 1827 ~~~~~~~~~~~~~ 1828 pos: 0 1829 flags: 04002 1830 mnt_id: 9 1831 eventfd-count: 5a 1832 1833 where 'eventfd-count' is hex value of a counter. 1834 1835 Signalfd files 1836 ~~~~~~~~~~~~~~ 1837 pos: 0 1838 flags: 04002 1839 mnt_id: 9 1840 sigmask: 0000000000000200 1841 1842 where 'sigmask' is hex value of the signal mask associated 1843 with a file. 1844 1845 Epoll files 1846 ~~~~~~~~~~~ 1847 pos: 0 1848 flags: 02 1849 mnt_id: 9 1850 tfd: 5 events: 1d data: ffffffffffffffff pos:0 ino:61af sdev:7 1851 1852 where 'tfd' is a target file descriptor number in decimal form, 1853 'events' is events mask being watched and the 'data' is data 1854 associated with a target [see epoll(7) for more details]. 1855 1856 The 'pos' is current offset of the target file in decimal form 1857 [see lseek(2)], 'ino' and 'sdev' are inode and device numbers 1858 where target file resides, all in hex format. 1859 1860 Fsnotify files 1861 ~~~~~~~~~~~~~~ 1862 For inotify files the format is the following 1863 1864 pos: 0 1865 flags: 02000000 1866 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d 1867 1868 where 'wd' is a watch descriptor in decimal form, ie a target file 1869 descriptor number, 'ino' and 'sdev' are inode and device where the 1870 target file resides and the 'mask' is the mask of events, all in hex 1871 form [see inotify(7) for more details]. 1872 1873 If the kernel was built with exportfs support, the path to the target 1874 file is encoded as a file handle. The file handle is provided by three 1875 fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex 1876 format. 1877 1878 If the kernel is built without exportfs support the file handle won't be 1879 printed out. 1880 1881 If there is no inotify mark attached yet the 'inotify' line will be omitted. 1882 1883 For fanotify files the format is 1884 1885 pos: 0 1886 flags: 02 1887 mnt_id: 9 1888 fanotify flags:10 event-flags:0 1889 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003 1890 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4 1891 1892 where fanotify 'flags' and 'event-flags' are values used in fanotify_init 1893 call, 'mnt_id' is the mount point identifier, 'mflags' is the value of 1894 flags associated with mark which are tracked separately from events 1895 mask. 'ino', 'sdev' are target inode and device, 'mask' is the events 1896 mask and 'ignored_mask' is the mask of events which are to be ignored. 1897 All in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask' 1898 does provide information about flags and mask used in fanotify_mark 1899 call [see fsnotify manpage for details]. 1900 1901 While the first three lines are mandatory and always printed, the rest is 1902 optional and may be omitted if no marks created yet. 1903 1904 Timerfd files 1905 ~~~~~~~~~~~~~ 1906 1907 pos: 0 1908 flags: 02 1909 mnt_id: 9 1910 clockid: 0 1911 ticks: 0 1912 settime flags: 01 1913 it_value: (0, 49406829) 1914 it_interval: (1, 0) 1915 1916 where 'clockid' is the clock type and 'ticks' is the number of the timer expirations 1917 that have occurred [see timerfd_create(2) for details]. 'settime flags' are 1918 flags in octal form been used to setup the timer [see timerfd_settime(2) for 1919 details]. 'it_value' is remaining time until the timer exiration. 1920 'it_interval' is the interval for the timer. Note the timer might be set up 1921 with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value' 1922 still exhibits timer's remaining time. 1923 19243.9 /proc/<pid>/map_files - Information about memory mapped files 1925--------------------------------------------------------------------- 1926This directory contains symbolic links which represent memory mapped files 1927the process is maintaining. Example output: 1928 1929 | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so 1930 | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so 1931 | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so 1932 | ... 1933 | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1 1934 | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls 1935 1936The name of a link represents the virtual memory bounds of a mapping, i.e. 1937vm_area_struct::vm_start-vm_area_struct::vm_end. 1938 1939The main purpose of the map_files is to retrieve a set of memory mapped 1940files in a fast way instead of parsing /proc/<pid>/maps or 1941/proc/<pid>/smaps, both of which contain many more records. At the same 1942time one can open(2) mappings from the listings of two processes and 1943comparing their inode numbers to figure out which anonymous memory areas 1944are actually shared. 1945 19463.10 /proc/<pid>/timerslack_ns - Task timerslack value 1947--------------------------------------------------------- 1948This file provides the value of the task's timerslack value in nanoseconds. 1949This value specifies a amount of time that normal timers may be deferred 1950in order to coalesce timers and avoid unnecessary wakeups. 1951 1952This allows a task's interactivity vs power consumption trade off to be 1953adjusted. 1954 1955Writing 0 to the file will set the tasks timerslack to the default value. 1956 1957Valid values are from 0 - ULLONG_MAX 1958 1959An application setting the value must have PTRACE_MODE_ATTACH_FSCREDS level 1960permissions on the task specified to change its timerslack_ns value. 1961 19623.11 /proc/<pid>/patch_state - Livepatch patch operation state 1963----------------------------------------------------------------- 1964When CONFIG_LIVEPATCH is enabled, this file displays the value of the 1965patch state for the task. 1966 1967A value of '-1' indicates that no patch is in transition. 1968 1969A value of '0' indicates that a patch is in transition and the task is 1970unpatched. If the patch is being enabled, then the task hasn't been 1971patched yet. If the patch is being disabled, then the task has already 1972been unpatched. 1973 1974A value of '1' indicates that a patch is in transition and the task is 1975patched. If the patch is being enabled, then the task has already been 1976patched. If the patch is being disabled, then the task hasn't been 1977unpatched yet. 1978 19793.12 /proc/<pid>/arch_status - task architecture specific status 1980------------------------------------------------------------------- 1981When CONFIG_PROC_PID_ARCH_STATUS is enabled, this file displays the 1982architecture specific status of the task. 1983 1984Example 1985------- 1986 $ cat /proc/6753/arch_status 1987 AVX512_elapsed_ms: 8 1988 1989Description 1990----------- 1991 1992x86 specific entries: 1993--------------------- 1994 AVX512_elapsed_ms: 1995 ------------------ 1996 If AVX512 is supported on the machine, this entry shows the milliseconds 1997 elapsed since the last time AVX512 usage was recorded. The recording 1998 happens on a best effort basis when a task is scheduled out. This means 1999 that the value depends on two factors: 2000 2001 1) The time which the task spent on the CPU without being scheduled 2002 out. With CPU isolation and a single runnable task this can take 2003 several seconds. 2004 2005 2) The time since the task was scheduled out last. Depending on the 2006 reason for being scheduled out (time slice exhausted, syscall ...) 2007 this can be arbitrary long time. 2008 2009 As a consequence the value cannot be considered precise and authoritative 2010 information. The application which uses this information has to be aware 2011 of the overall scenario on the system in order to determine whether a 2012 task is a real AVX512 user or not. Precise information can be obtained 2013 with performance counters. 2014 2015 A special value of '-1' indicates that no AVX512 usage was recorded, thus 2016 the task is unlikely an AVX512 user, but depends on the workload and the 2017 scheduling scenario, it also could be a false negative mentioned above. 2018 2019------------------------------------------------------------------------------ 2020Configuring procfs 2021------------------------------------------------------------------------------ 2022 20234.1 Mount options 2024--------------------- 2025 2026The following mount options are supported: 2027 2028 hidepid= Set /proc/<pid>/ access mode. 2029 gid= Set the group authorized to learn processes information. 2030 2031hidepid=0 means classic mode - everybody may access all /proc/<pid>/ directories 2032(default). 2033 2034hidepid=1 means users may not access any /proc/<pid>/ directories but their 2035own. Sensitive files like cmdline, sched*, status are now protected against 2036other users. This makes it impossible to learn whether any user runs 2037specific program (given the program doesn't reveal itself by its behaviour). 2038As an additional bonus, as /proc/<pid>/cmdline is unaccessible for other users, 2039poorly written programs passing sensitive information via program arguments are 2040now protected against local eavesdroppers. 2041 2042hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be fully invisible to other 2043users. It doesn't mean that it hides a fact whether a process with a specific 2044pid value exists (it can be learned by other means, e.g. by "kill -0 $PID"), 2045but it hides process' uid and gid, which may be learned by stat()'ing 2046/proc/<pid>/ otherwise. It greatly complicates an intruder's task of gathering 2047information about running processes, whether some daemon runs with elevated 2048privileges, whether other user runs some sensitive program, whether other users 2049run any program at all, etc. 2050 2051gid= defines a group authorized to learn processes information otherwise 2052prohibited by hidepid=. If you use some daemon like identd which needs to learn 2053information about processes information, just add identd to this group. 2054