| xj | b04a402 | 2021-11-25 15:01:52 +0800 | [diff] [blame] | 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 | |
| 7 | 2.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000 |
| 8 | move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009 |
| 9 | ------------------------------------------------------------------------------ |
| 10 | Version 1.3 Kernel version 2.2.12 |
| 11 | Kernel version 2.4.0-test11-pre4 |
| 12 | ------------------------------------------------------------------------------ |
| 13 | fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009 |
| 14 | |
| 15 | Table 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 | |
| 49 | 4 Configuring procfs |
| 50 | 4.1 Mount options |
| 51 | |
| 52 | ------------------------------------------------------------------------------ |
| 53 | Preface |
| 54 | ------------------------------------------------------------------------------ |
| 55 | |
| 56 | 0.1 Introduction/Credits |
| 57 | ------------------------ |
| 58 | |
| 59 | This documentation is part of a soon (or so we hope) to be released book on |
| 60 | the SuSE Linux distribution. As there is no complete documentation for the |
| 61 | /proc file system and we've used many freely available sources to write these |
| 62 | chapters, it seems only fair to give the work back to the Linux community. |
| 63 | This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm |
| 64 | afraid it's still far from complete, but we hope it will be useful. As far as |
| 65 | we know, it is the first 'all-in-one' document about the /proc file system. It |
| 66 | is focused on the Intel x86 hardware, so if you are looking for PPC, ARM, |
| 67 | SPARC, AXP, etc., features, you probably won't find what you are looking for. |
| 68 | It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But |
| 69 | additions and patches are welcome and will be added to this document if you |
| 70 | mail them to Bodo. |
| 71 | |
| 72 | We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of |
| 73 | other people for help compiling this documentation. We'd also like to extend a |
| 74 | special thank you to Andi Kleen for documentation, which we relied on heavily |
| 75 | to create this document, as well as the additional information he provided. |
| 76 | Thanks to everybody else who contributed source or docs to the Linux kernel |
| 77 | and helped create a great piece of software... :) |
| 78 | |
| 79 | If you have any comments, corrections or additions, please don't hesitate to |
| 80 | contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this |
| 81 | document. |
| 82 | |
| 83 | The latest version of this document is available online at |
| 84 | http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html |
| 85 | |
| 86 | If the above direction does not works for you, you could try the kernel |
| 87 | mailing list at linux-kernel@vger.kernel.org and/or try to reach me at |
| 88 | comandante@zaralinux.com. |
| 89 | |
| 90 | 0.2 Legal Stuff |
| 91 | --------------- |
| 92 | |
| 93 | We don't guarantee the correctness of this document, and if you come to us |
| 94 | complaining about how you screwed up your system because of incorrect |
| 95 | documentation, we won't feel responsible... |
| 96 | |
| 97 | ------------------------------------------------------------------------------ |
| 98 | CHAPTER 1: COLLECTING SYSTEM INFORMATION |
| 99 | ------------------------------------------------------------------------------ |
| 100 | |
| 101 | ------------------------------------------------------------------------------ |
| 102 | In This Chapter |
| 103 | ------------------------------------------------------------------------------ |
| 104 | * Investigating the properties of the pseudo file system /proc and its |
| 105 | ability to provide information on the running Linux system |
| 106 | * Examining /proc's structure |
| 107 | * Uncovering various information about the kernel and the processes running |
| 108 | on the system |
| 109 | ------------------------------------------------------------------------------ |
| 110 | |
| 111 | |
| 112 | The proc file system acts as an interface to internal data structures in the |
| 113 | kernel. It can be used to obtain information about the system and to change |
| 114 | certain kernel parameters at runtime (sysctl). |
| 115 | |
| 116 | First, we'll take a look at the read-only parts of /proc. In Chapter 2, we |
| 117 | show you how you can use /proc/sys to change settings. |
| 118 | |
| 119 | 1.1 Process-Specific Subdirectories |
| 120 | ----------------------------------- |
| 121 | |
| 122 | The directory /proc contains (among other things) one subdirectory for each |
| 123 | process running on the system, which is named after the process ID (PID). |
| 124 | |
| 125 | The link self points to the process reading the file system. Each process |
| 126 | subdirectory has the entries listed in Table 1-1. |
| 127 | |
| 128 | |
| 129 | Table 1-1: Process specific entries in /proc |
| 130 | .............................................................................. |
| 131 | File Content |
| 132 | clear_refs Clears page referenced bits shown in smaps output |
| 133 | cmdline Command line arguments |
| 134 | cpu Current and last cpu in which it was executed (2.4)(smp) |
| 135 | cwd Link to the current working directory |
| 136 | environ Values of environment variables |
| 137 | exe Link to the executable of this process |
| 138 | fd Directory, which contains all file descriptors |
| 139 | maps Memory maps to executables and library files (2.4) |
| 140 | mem Memory held by this process |
| 141 | root Link to the root directory of this process |
| 142 | stat Process status |
| 143 | statm Process memory status information |
| 144 | status Process status in human readable form |
| 145 | wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function |
| 146 | symbol the task is blocked in - or "0" if not blocked. |
| 147 | pagemap Page table |
| 148 | stack Report full stack trace, enable via CONFIG_STACKTRACE |
| 149 | smaps an extension based on maps, showing the memory consumption of |
| 150 | each mapping and flags associated with it |
| 151 | numa_maps an extension based on maps, showing the memory locality and |
| 152 | binding policy as well as mem usage (in pages) of each mapping. |
| 153 | .............................................................................. |
| 154 | |
| 155 | For example, to get the status information of a process, all you have to do is |
| 156 | read the file /proc/PID/status: |
| 157 | |
| 158 | >cat /proc/self/status |
| 159 | Name: cat |
| 160 | State: R (running) |
| 161 | Tgid: 5452 |
| 162 | Pid: 5452 |
| 163 | PPid: 743 |
| 164 | TracerPid: 0 (2.4) |
| 165 | Uid: 501 501 501 501 |
| 166 | Gid: 100 100 100 100 |
| 167 | FDSize: 256 |
| 168 | Groups: 100 14 16 |
| 169 | VmPeak: 5004 kB |
| 170 | VmSize: 5004 kB |
| 171 | VmLck: 0 kB |
| 172 | VmHWM: 476 kB |
| 173 | VmRSS: 476 kB |
| 174 | RssAnon: 352 kB |
| 175 | RssFile: 120 kB |
| 176 | RssShmem: 4 kB |
| 177 | VmData: 156 kB |
| 178 | VmStk: 88 kB |
| 179 | VmExe: 68 kB |
| 180 | VmLib: 1412 kB |
| 181 | VmPTE: 20 kb |
| 182 | VmSwap: 0 kB |
| 183 | HugetlbPages: 0 kB |
| 184 | CoreDumping: 0 |
| 185 | Threads: 1 |
| 186 | SigQ: 0/28578 |
| 187 | SigPnd: 0000000000000000 |
| 188 | ShdPnd: 0000000000000000 |
| 189 | SigBlk: 0000000000000000 |
| 190 | SigIgn: 0000000000000000 |
| 191 | SigCgt: 0000000000000000 |
| 192 | CapInh: 00000000fffffeff |
| 193 | CapPrm: 0000000000000000 |
| 194 | CapEff: 0000000000000000 |
| 195 | CapBnd: ffffffffffffffff |
| 196 | NoNewPrivs: 0 |
| 197 | Seccomp: 0 |
| 198 | voluntary_ctxt_switches: 0 |
| 199 | nonvoluntary_ctxt_switches: 1 |
| 200 | |
| 201 | This shows you nearly the same information you would get if you viewed it with |
| 202 | the ps command. In fact, ps uses the proc file system to obtain its |
| 203 | information. But you get a more detailed view of the process by reading the |
| 204 | file /proc/PID/status. It fields are described in table 1-2. |
| 205 | |
| 206 | The statm file contains more detailed information about the process |
| 207 | memory usage. Its seven fields are explained in Table 1-3. The stat file |
| 208 | contains details information about the process itself. Its fields are |
| 209 | explained in Table 1-4. |
| 210 | |
| 211 | (for SMP CONFIG users) |
| 212 | For making accounting scalable, RSS related information are handled in an |
| 213 | asynchronous manner and the value may not be very precise. To see a precise |
| 214 | snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table. |
| 215 | It's slow but very precise. |
| 216 | |
| 217 | Table 1-2: Contents of the status files (as of 4.8) |
| 218 | .............................................................................. |
| 219 | Field Content |
| 220 | Name filename of the executable |
| 221 | Umask file mode creation mask |
| 222 | State state (R is running, S is sleeping, D is sleeping |
| 223 | in an uninterruptible wait, Z is zombie, |
| 224 | T is traced or stopped) |
| 225 | Tgid thread group ID |
| 226 | Ngid NUMA group ID (0 if none) |
| 227 | Pid process id |
| 228 | PPid process id of the parent process |
| 229 | TracerPid PID of process tracing this process (0 if not) |
| 230 | Uid Real, effective, saved set, and file system UIDs |
| 231 | Gid Real, effective, saved set, and file system GIDs |
| 232 | FDSize number of file descriptor slots currently allocated |
| 233 | Groups supplementary group list |
| 234 | NStgid descendant namespace thread group ID hierarchy |
| 235 | NSpid descendant namespace process ID hierarchy |
| 236 | NSpgid descendant namespace process group ID hierarchy |
| 237 | NSsid descendant namespace session ID hierarchy |
| 238 | VmPeak peak virtual memory size |
| 239 | VmSize total program size |
| 240 | VmLck locked memory size |
| 241 | VmPin pinned memory size |
| 242 | VmHWM peak resident set size ("high water mark") |
| 243 | VmRSS size of memory portions. It contains the three |
| 244 | following parts (VmRSS = RssAnon + RssFile + RssShmem) |
| 245 | RssAnon size of resident anonymous memory |
| 246 | RssFile size of resident file mappings |
| 247 | RssShmem size of resident shmem memory (includes SysV shm, |
| 248 | mapping of tmpfs and shared anonymous mappings) |
| 249 | VmData size of private data segments |
| 250 | VmStk size of stack segments |
| 251 | VmExe size of text segment |
| 252 | VmLib size of shared library code |
| 253 | VmPTE size of page table entries |
| 254 | VmSwap amount of swap used by anonymous private data |
| 255 | (shmem swap usage is not included) |
| 256 | HugetlbPages size of hugetlb memory portions |
| 257 | CoreDumping process's memory is currently being dumped |
| 258 | (killing the process may lead to a corrupted core) |
| 259 | Threads number of threads |
| 260 | SigQ number of signals queued/max. number for queue |
| 261 | SigPnd bitmap of pending signals for the thread |
| 262 | ShdPnd bitmap of shared pending signals for the process |
| 263 | SigBlk bitmap of blocked signals |
| 264 | SigIgn bitmap of ignored signals |
| 265 | SigCgt bitmap of caught signals |
| 266 | CapInh bitmap of inheritable capabilities |
| 267 | CapPrm bitmap of permitted capabilities |
| 268 | CapEff bitmap of effective capabilities |
| 269 | CapBnd bitmap of capabilities bounding set |
| 270 | NoNewPrivs no_new_privs, like prctl(PR_GET_NO_NEW_PRIV, ...) |
| 271 | Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...) |
| 272 | Cpus_allowed mask of CPUs on which this process may run |
| 273 | Cpus_allowed_list Same as previous, but in "list format" |
| 274 | Mems_allowed mask of memory nodes allowed to this process |
| 275 | Mems_allowed_list Same as previous, but in "list format" |
| 276 | voluntary_ctxt_switches number of voluntary context switches |
| 277 | nonvoluntary_ctxt_switches number of non voluntary context switches |
| 278 | .............................................................................. |
| 279 | |
| 280 | Table 1-3: Contents of the statm files (as of 2.6.8-rc3) |
| 281 | .............................................................................. |
| 282 | Field Content |
| 283 | size total program size (pages) (same as VmSize in status) |
| 284 | resident size of memory portions (pages) (same as VmRSS in status) |
| 285 | shared number of pages that are shared (i.e. backed by a file, same |
| 286 | as RssFile+RssShmem in status) |
| 287 | trs number of pages that are 'code' (not including libs; broken, |
| 288 | includes data segment) |
| 289 | lrs number of pages of library (always 0 on 2.6) |
| 290 | drs number of pages of data/stack (including libs; broken, |
| 291 | includes library text) |
| 292 | dt number of dirty pages (always 0 on 2.6) |
| 293 | .............................................................................. |
| 294 | |
| 295 | |
| 296 | Table 1-4: Contents of the stat files (as of 2.6.30-rc7) |
| 297 | .............................................................................. |
| 298 | Field Content |
| 299 | pid process id |
| 300 | tcomm filename of the executable |
| 301 | state state (R is running, S is sleeping, D is sleeping in an |
| 302 | uninterruptible wait, Z is zombie, T is traced or stopped) |
| 303 | ppid process id of the parent process |
| 304 | pgrp pgrp of the process |
| 305 | sid session id |
| 306 | tty_nr tty the process uses |
| 307 | tty_pgrp pgrp of the tty |
| 308 | flags task flags |
| 309 | min_flt number of minor faults |
| 310 | cmin_flt number of minor faults with child's |
| 311 | maj_flt number of major faults |
| 312 | cmaj_flt number of major faults with child's |
| 313 | utime user mode jiffies |
| 314 | stime kernel mode jiffies |
| 315 | cutime user mode jiffies with child's |
| 316 | cstime kernel mode jiffies with child's |
| 317 | priority priority level |
| 318 | nice nice level |
| 319 | num_threads number of threads |
| 320 | it_real_value (obsolete, always 0) |
| 321 | start_time time the process started after system boot |
| 322 | vsize virtual memory size |
| 323 | rss resident set memory size |
| 324 | rsslim current limit in bytes on the rss |
| 325 | start_code address above which program text can run |
| 326 | end_code address below which program text can run |
| 327 | start_stack address of the start of the main process stack |
| 328 | esp current value of ESP |
| 329 | eip current value of EIP |
| 330 | pending bitmap of pending signals |
| 331 | blocked bitmap of blocked signals |
| 332 | sigign bitmap of ignored signals |
| 333 | sigcatch bitmap of caught signals |
| 334 | 0 (place holder, used to be the wchan address, use /proc/PID/wchan instead) |
| 335 | 0 (place holder) |
| 336 | 0 (place holder) |
| 337 | exit_signal signal to send to parent thread on exit |
| 338 | task_cpu which CPU the task is scheduled on |
| 339 | rt_priority realtime priority |
| 340 | policy scheduling policy (man sched_setscheduler) |
| 341 | blkio_ticks time spent waiting for block IO |
| 342 | gtime guest time of the task in jiffies |
| 343 | cgtime guest time of the task children in jiffies |
| 344 | start_data address above which program data+bss is placed |
| 345 | end_data address below which program data+bss is placed |
| 346 | start_brk address above which program heap can be expanded with brk() |
| 347 | arg_start address above which program command line is placed |
| 348 | arg_end address below which program command line is placed |
| 349 | env_start address above which program environment is placed |
| 350 | env_end address below which program environment is placed |
| 351 | exit_code the thread's exit_code in the form reported by the waitpid system call |
| 352 | .............................................................................. |
| 353 | |
| 354 | The /proc/PID/maps file containing the currently mapped memory regions and |
| 355 | their access permissions. |
| 356 | |
| 357 | The format is: |
| 358 | |
| 359 | address perms offset dev inode pathname |
| 360 | |
| 361 | 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test |
| 362 | 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test |
| 363 | 0804a000-0806b000 rw-p 00000000 00:00 0 [heap] |
| 364 | a7cb1000-a7cb2000 ---p 00000000 00:00 0 |
| 365 | a7cb2000-a7eb2000 rw-p 00000000 00:00 0 |
| 366 | a7eb2000-a7eb3000 ---p 00000000 00:00 0 |
| 367 | a7eb3000-a7ed5000 rw-p 00000000 00:00 0 |
| 368 | a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6 |
| 369 | a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6 |
| 370 | a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6 |
| 371 | a800b000-a800e000 rw-p 00000000 00:00 0 |
| 372 | a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0 |
| 373 | a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0 |
| 374 | a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0 |
| 375 | a8024000-a8027000 rw-p 00000000 00:00 0 |
| 376 | a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2 |
| 377 | a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2 |
| 378 | a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2 |
| 379 | aff35000-aff4a000 rw-p 00000000 00:00 0 [stack] |
| 380 | ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso] |
| 381 | |
| 382 | where "address" is the address space in the process that it occupies, "perms" |
| 383 | is a set of permissions: |
| 384 | |
| 385 | r = read |
| 386 | w = write |
| 387 | x = execute |
| 388 | s = shared |
| 389 | p = private (copy on write) |
| 390 | |
| 391 | "offset" is the offset into the mapping, "dev" is the device (major:minor), and |
| 392 | "inode" is the inode on that device. 0 indicates that no inode is associated |
| 393 | with the memory region, as the case would be with BSS (uninitialized data). |
| 394 | The "pathname" shows the name associated file for this mapping. If the mapping |
| 395 | is not associated with a file: |
| 396 | |
| 397 | [heap] = the heap of the program |
| 398 | [stack] = the stack of the main process |
| 399 | [vdso] = the "virtual dynamic shared object", |
| 400 | the kernel system call handler |
| 401 | [anon:<name>] = an anonymous mapping that has been |
| 402 | named by userspace |
| 403 | |
| 404 | or if empty, the mapping is anonymous. |
| 405 | |
| 406 | The /proc/PID/smaps is an extension based on maps, showing the memory |
| 407 | consumption for each of the process's mappings. For each of mappings there |
| 408 | is a series of lines such as the following: |
| 409 | |
| 410 | 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash |
| 411 | Size: 1084 kB |
| 412 | Rss: 892 kB |
| 413 | Pss: 374 kB |
| 414 | Shared_Clean: 892 kB |
| 415 | Shared_Dirty: 0 kB |
| 416 | Private_Clean: 0 kB |
| 417 | Private_Dirty: 0 kB |
| 418 | Referenced: 892 kB |
| 419 | Anonymous: 0 kB |
| 420 | LazyFree: 0 kB |
| 421 | AnonHugePages: 0 kB |
| 422 | ShmemPmdMapped: 0 kB |
| 423 | Shared_Hugetlb: 0 kB |
| 424 | Private_Hugetlb: 0 kB |
| 425 | Swap: 0 kB |
| 426 | SwapPss: 0 kB |
| 427 | KernelPageSize: 4 kB |
| 428 | MMUPageSize: 4 kB |
| 429 | Locked: 0 kB |
| 430 | THPeligible: 0 |
| 431 | VmFlags: rd ex mr mw me dw |
| 432 | Name: name from userspace |
| 433 | |
| 434 | the first of these lines shows the same information as is displayed for the |
| 435 | mapping in /proc/PID/maps. The remaining lines show the size of the mapping |
| 436 | (size), the amount of the mapping that is currently resident in RAM (RSS), the |
| 437 | process' proportional share of this mapping (PSS), the number of clean and |
| 438 | dirty private pages in the mapping. |
| 439 | |
| 440 | The "proportional set size" (PSS) of a process is the count of pages it has |
| 441 | in memory, where each page is divided by the number of processes sharing it. |
| 442 | So if a process has 1000 pages all to itself, and 1000 shared with one other |
| 443 | process, its PSS will be 1500. |
| 444 | Note that even a page which is part of a MAP_SHARED mapping, but has only |
| 445 | a single pte mapped, i.e. is currently used by only one process, is accounted |
| 446 | as private and not as shared. |
| 447 | "Referenced" indicates the amount of memory currently marked as referenced or |
| 448 | accessed. |
| 449 | "Anonymous" shows the amount of memory that does not belong to any file. Even |
| 450 | a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE |
| 451 | and a page is modified, the file page is replaced by a private anonymous copy. |
| 452 | "LazyFree" shows the amount of memory which is marked by madvise(MADV_FREE). |
| 453 | The memory isn't freed immediately with madvise(). It's freed in memory |
| 454 | pressure if the memory is clean. Please note that the printed value might |
| 455 | be lower than the real value due to optimizations used in the current |
| 456 | implementation. If this is not desirable please file a bug report. |
| 457 | "AnonHugePages" shows the ammount of memory backed by transparent hugepage. |
| 458 | "ShmemPmdMapped" shows the ammount of shared (shmem/tmpfs) memory backed by |
| 459 | huge pages. |
| 460 | "Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by |
| 461 | hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical |
| 462 | reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field. |
| 463 | "Swap" shows how much would-be-anonymous memory is also used, but out on swap. |
| 464 | For shmem mappings, "Swap" includes also the size of the mapped (and not |
| 465 | replaced by copy-on-write) part of the underlying shmem object out on swap. |
| 466 | "SwapPss" shows proportional swap share of this mapping. Unlike "Swap", this |
| 467 | does not take into account swapped out page of underlying shmem objects. |
| 468 | "Locked" indicates whether the mapping is locked in memory or not. |
| 469 | "THPeligible" indicates whether the mapping is eligible for THP pages - 1 if |
| 470 | true, 0 otherwise. |
| 471 | |
| 472 | "VmFlags" field deserves a separate description. This member represents the kernel |
| 473 | flags associated with the particular virtual memory area in two letter encoded |
| 474 | manner. The codes are the following: |
| 475 | rd - readable |
| 476 | wr - writeable |
| 477 | ex - executable |
| 478 | sh - shared |
| 479 | mr - may read |
| 480 | mw - may write |
| 481 | me - may execute |
| 482 | ms - may share |
| 483 | gd - stack segment growns down |
| 484 | pf - pure PFN range |
| 485 | dw - disabled write to the mapped file |
| 486 | lo - pages are locked in memory |
| 487 | io - memory mapped I/O area |
| 488 | sr - sequential read advise provided |
| 489 | rr - random read advise provided |
| 490 | dc - do not copy area on fork |
| 491 | de - do not expand area on remapping |
| 492 | ac - area is accountable |
| 493 | nr - swap space is not reserved for the area |
| 494 | ht - area uses huge tlb pages |
| 495 | ar - architecture specific flag |
| 496 | dd - do not include area into core dump |
| 497 | sd - soft-dirty flag |
| 498 | mm - mixed map area |
| 499 | hg - huge page advise flag |
| 500 | nh - no-huge page advise flag |
| 501 | mg - mergable advise flag |
| 502 | |
| 503 | Note that there is no guarantee that every flag and associated mnemonic will |
| 504 | be present in all further kernel releases. Things get changed, the flags may |
| 505 | be vanished or the reverse -- new added. Interpretation of their meaning |
| 506 | might change in future as well. So each consumer of these flags has to |
| 507 | follow each specific kernel version for the exact semantic. |
| 508 | |
| 509 | The "Name" field will only be present on a mapping that has been named by |
| 510 | userspace, and will show the name passed in by userspace. |
| 511 | |
| 512 | This file is only present if the CONFIG_MMU kernel configuration option is |
| 513 | enabled. |
| 514 | |
| 515 | Note: reading /proc/PID/maps or /proc/PID/smaps is inherently racy (consistent |
| 516 | output can be achieved only in the single read call). |
| 517 | This typically manifests when doing partial reads of these files while the |
| 518 | memory map is being modified. Despite the races, we do provide the following |
| 519 | guarantees: |
| 520 | |
| 521 | 1) The mapped addresses never go backwards, which implies no two |
| 522 | regions will ever overlap. |
| 523 | 2) If there is something at a given vaddr during the entirety of the |
| 524 | life of the smaps/maps walk, there will be some output for it. |
| 525 | |
| 526 | |
| 527 | The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG |
| 528 | bits on both physical and virtual pages associated with a process, and the |
| 529 | soft-dirty bit on pte (see Documentation/admin-guide/mm/soft-dirty.rst |
| 530 | for details). |
| 531 | To clear the bits for all the pages associated with the process |
| 532 | > echo 1 > /proc/PID/clear_refs |
| 533 | |
| 534 | To clear the bits for the anonymous pages associated with the process |
| 535 | > echo 2 > /proc/PID/clear_refs |
| 536 | |
| 537 | To clear the bits for the file mapped pages associated with the process |
| 538 | > echo 3 > /proc/PID/clear_refs |
| 539 | |
| 540 | To clear the soft-dirty bit |
| 541 | > echo 4 > /proc/PID/clear_refs |
| 542 | |
| 543 | To reset the peak resident set size ("high water mark") to the process's |
| 544 | current value: |
| 545 | > echo 5 > /proc/PID/clear_refs |
| 546 | |
| 547 | Any other value written to /proc/PID/clear_refs will have no effect. |
| 548 | |
| 549 | The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags |
| 550 | using /proc/kpageflags and number of times a page is mapped using |
| 551 | /proc/kpagecount. For detailed explanation, see |
| 552 | Documentation/admin-guide/mm/pagemap.rst. |
| 553 | |
| 554 | The /proc/pid/numa_maps is an extension based on maps, showing the memory |
| 555 | locality and binding policy, as well as the memory usage (in pages) of |
| 556 | each mapping. The output follows a general format where mapping details get |
| 557 | summarized separated by blank spaces, one mapping per each file line: |
| 558 | |
| 559 | address policy mapping details |
| 560 | |
| 561 | 00400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4 |
| 562 | 00600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4 |
| 563 | 3206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4 |
| 564 | 320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 |
| 565 | 3206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 |
| 566 | 3206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4 |
| 567 | 3206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4 |
| 568 | 320698b000 default file=/lib64/libc-2.12.so |
| 569 | 3206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4 |
| 570 | 3206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 |
| 571 | 3206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4 |
| 572 | 7f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4 |
| 573 | 7f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4 |
| 574 | 7f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048 |
| 575 | 7fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4 |
| 576 | 7fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4 |
| 577 | |
| 578 | Where: |
| 579 | "address" is the starting address for the mapping; |
| 580 | "policy" reports the NUMA memory policy set for the mapping (see Documentation/admin-guide/mm/numa_memory_policy.rst); |
| 581 | "mapping details" summarizes mapping data such as mapping type, page usage counters, |
| 582 | node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page |
| 583 | size, in KB, that is backing the mapping up. |
| 584 | |
| 585 | 1.2 Kernel data |
| 586 | --------------- |
| 587 | |
| 588 | Similar to the process entries, the kernel data files give information about |
| 589 | the running kernel. The files used to obtain this information are contained in |
| 590 | /proc and are listed in Table 1-5. Not all of these will be present in your |
| 591 | system. It depends on the kernel configuration and the loaded modules, which |
| 592 | files are there, and which are missing. |
| 593 | |
| 594 | Table 1-5: Kernel info in /proc |
| 595 | .............................................................................. |
| 596 | File Content |
| 597 | apm Advanced power management info |
| 598 | buddyinfo Kernel memory allocator information (see text) (2.5) |
| 599 | bus Directory containing bus specific information |
| 600 | cmdline Kernel command line |
| 601 | cpuinfo Info about the CPU |
| 602 | devices Available devices (block and character) |
| 603 | dma Used DMS channels |
| 604 | filesystems Supported filesystems |
| 605 | driver Various drivers grouped here, currently rtc (2.4) |
| 606 | execdomains Execdomains, related to security (2.4) |
| 607 | fb Frame Buffer devices (2.4) |
| 608 | fs File system parameters, currently nfs/exports (2.4) |
| 609 | ide Directory containing info about the IDE subsystem |
| 610 | interrupts Interrupt usage |
| 611 | iomem Memory map (2.4) |
| 612 | ioports I/O port usage |
| 613 | irq Masks for irq to cpu affinity (2.4)(smp?) |
| 614 | isapnp ISA PnP (Plug&Play) Info (2.4) |
| 615 | kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4)) |
| 616 | kmsg Kernel messages |
| 617 | ksyms Kernel symbol table |
| 618 | loadavg Load average of last 1, 5 & 15 minutes |
| 619 | locks Kernel locks |
| 620 | meminfo Memory info |
| 621 | misc Miscellaneous |
| 622 | modules List of loaded modules |
| 623 | mounts Mounted filesystems |
| 624 | net Networking info (see text) |
| 625 | pagetypeinfo Additional page allocator information (see text) (2.5) |
| 626 | partitions Table of partitions known to the system |
| 627 | pci Deprecated info of PCI bus (new way -> /proc/bus/pci/, |
| 628 | decoupled by lspci (2.4) |
| 629 | rtc Real time clock |
| 630 | scsi SCSI info (see text) |
| 631 | slabinfo Slab pool info |
| 632 | softirqs softirq usage |
| 633 | stat Overall statistics |
| 634 | swaps Swap space utilization |
| 635 | sys See chapter 2 |
| 636 | sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4) |
| 637 | tty Info of tty drivers |
| 638 | uptime Wall clock since boot, combined idle time of all cpus |
| 639 | version Kernel version |
| 640 | video bttv info of video resources (2.4) |
| 641 | vmallocinfo Show vmalloced areas |
| 642 | .............................................................................. |
| 643 | |
| 644 | You can, for example, check which interrupts are currently in use and what |
| 645 | they are used for by looking in the file /proc/interrupts: |
| 646 | |
| 647 | > cat /proc/interrupts |
| 648 | CPU0 |
| 649 | 0: 8728810 XT-PIC timer |
| 650 | 1: 895 XT-PIC keyboard |
| 651 | 2: 0 XT-PIC cascade |
| 652 | 3: 531695 XT-PIC aha152x |
| 653 | 4: 2014133 XT-PIC serial |
| 654 | 5: 44401 XT-PIC pcnet_cs |
| 655 | 8: 2 XT-PIC rtc |
| 656 | 11: 8 XT-PIC i82365 |
| 657 | 12: 182918 XT-PIC PS/2 Mouse |
| 658 | 13: 1 XT-PIC fpu |
| 659 | 14: 1232265 XT-PIC ide0 |
| 660 | 15: 7 XT-PIC ide1 |
| 661 | NMI: 0 |
| 662 | |
| 663 | In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the |
| 664 | output of a SMP machine): |
| 665 | |
| 666 | > cat /proc/interrupts |
| 667 | |
| 668 | CPU0 CPU1 |
| 669 | 0: 1243498 1214548 IO-APIC-edge timer |
| 670 | 1: 8949 8958 IO-APIC-edge keyboard |
| 671 | 2: 0 0 XT-PIC cascade |
| 672 | 5: 11286 10161 IO-APIC-edge soundblaster |
| 673 | 8: 1 0 IO-APIC-edge rtc |
| 674 | 9: 27422 27407 IO-APIC-edge 3c503 |
| 675 | 12: 113645 113873 IO-APIC-edge PS/2 Mouse |
| 676 | 13: 0 0 XT-PIC fpu |
| 677 | 14: 22491 24012 IO-APIC-edge ide0 |
| 678 | 15: 2183 2415 IO-APIC-edge ide1 |
| 679 | 17: 30564 30414 IO-APIC-level eth0 |
| 680 | 18: 177 164 IO-APIC-level bttv |
| 681 | NMI: 2457961 2457959 |
| 682 | LOC: 2457882 2457881 |
| 683 | ERR: 2155 |
| 684 | |
| 685 | NMI is incremented in this case because every timer interrupt generates a NMI |
| 686 | (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups. |
| 687 | |
| 688 | LOC is the local interrupt counter of the internal APIC of every CPU. |
| 689 | |
| 690 | ERR is incremented in the case of errors in the IO-APIC bus (the bus that |
| 691 | connects the CPUs in a SMP system. This means that an error has been detected, |
| 692 | the IO-APIC automatically retry the transmission, so it should not be a big |
| 693 | problem, but you should read the SMP-FAQ. |
| 694 | |
| 695 | In 2.6.2* /proc/interrupts was expanded again. This time the goal was for |
| 696 | /proc/interrupts to display every IRQ vector in use by the system, not |
| 697 | just those considered 'most important'. The new vectors are: |
| 698 | |
| 699 | THR -- interrupt raised when a machine check threshold counter |
| 700 | (typically counting ECC corrected errors of memory or cache) exceeds |
| 701 | a configurable threshold. Only available on some systems. |
| 702 | |
| 703 | TRM -- a thermal event interrupt occurs when a temperature threshold |
| 704 | has been exceeded for the CPU. This interrupt may also be generated |
| 705 | when the temperature drops back to normal. |
| 706 | |
| 707 | SPU -- a spurious interrupt is some interrupt that was raised then lowered |
| 708 | by some IO device before it could be fully processed by the APIC. Hence |
| 709 | the APIC sees the interrupt but does not know what device it came from. |
| 710 | For this case the APIC will generate the interrupt with a IRQ vector |
| 711 | of 0xff. This might also be generated by chipset bugs. |
| 712 | |
| 713 | RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are |
| 714 | sent from one CPU to another per the needs of the OS. Typically, |
| 715 | their statistics are used by kernel developers and interested users to |
| 716 | determine the occurrence of interrupts of the given type. |
| 717 | |
| 718 | The above IRQ vectors are displayed only when relevant. For example, |
| 719 | the threshold vector does not exist on x86_64 platforms. Others are |
| 720 | suppressed when the system is a uniprocessor. As of this writing, only |
| 721 | i386 and x86_64 platforms support the new IRQ vector displays. |
| 722 | |
| 723 | Of some interest is the introduction of the /proc/irq directory to 2.4. |
| 724 | It could be used to set IRQ to CPU affinity, this means that you can "hook" an |
| 725 | IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the |
| 726 | irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and |
| 727 | prof_cpu_mask. |
| 728 | |
| 729 | For example |
| 730 | > ls /proc/irq/ |
| 731 | 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask |
| 732 | 1 11 13 15 17 19 3 5 7 9 default_smp_affinity |
| 733 | > ls /proc/irq/0/ |
| 734 | smp_affinity |
| 735 | |
| 736 | smp_affinity is a bitmask, in which you can specify which CPUs can handle the |
| 737 | IRQ, you can set it by doing: |
| 738 | |
| 739 | > echo 1 > /proc/irq/10/smp_affinity |
| 740 | |
| 741 | This means that only the first CPU will handle the IRQ, but you can also echo |
| 742 | 5 which means that only the first and third CPU can handle the IRQ. |
| 743 | |
| 744 | The contents of each smp_affinity file is the same by default: |
| 745 | |
| 746 | > cat /proc/irq/0/smp_affinity |
| 747 | ffffffff |
| 748 | |
| 749 | There is an alternate interface, smp_affinity_list which allows specifying |
| 750 | a cpu range instead of a bitmask: |
| 751 | |
| 752 | > cat /proc/irq/0/smp_affinity_list |
| 753 | 1024-1031 |
| 754 | |
| 755 | The default_smp_affinity mask applies to all non-active IRQs, which are the |
| 756 | IRQs which have not yet been allocated/activated, and hence which lack a |
| 757 | /proc/irq/[0-9]* directory. |
| 758 | |
| 759 | The node file on an SMP system shows the node to which the device using the IRQ |
| 760 | reports itself as being attached. This hardware locality information does not |
| 761 | include information about any possible driver locality preference. |
| 762 | |
| 763 | prof_cpu_mask specifies which CPUs are to be profiled by the system wide |
| 764 | profiler. Default value is ffffffff (all cpus if there are only 32 of them). |
| 765 | |
| 766 | The way IRQs are routed is handled by the IO-APIC, and it's Round Robin |
| 767 | between all the CPUs which are allowed to handle it. As usual the kernel has |
| 768 | more info than you and does a better job than you, so the defaults are the |
| 769 | best choice for almost everyone. [Note this applies only to those IO-APIC's |
| 770 | that support "Round Robin" interrupt distribution.] |
| 771 | |
| 772 | There are three more important subdirectories in /proc: net, scsi, and sys. |
| 773 | The general rule is that the contents, or even the existence of these |
| 774 | directories, depend on your kernel configuration. If SCSI is not enabled, the |
| 775 | directory scsi may not exist. The same is true with the net, which is there |
| 776 | only when networking support is present in the running kernel. |
| 777 | |
| 778 | The slabinfo file gives information about memory usage at the slab level. |
| 779 | Linux uses slab pools for memory management above page level in version 2.2. |
| 780 | Commonly used objects have their own slab pool (such as network buffers, |
| 781 | directory cache, and so on). |
| 782 | |
| 783 | .............................................................................. |
| 784 | |
| 785 | > cat /proc/buddyinfo |
| 786 | |
| 787 | Node 0, zone DMA 0 4 5 4 4 3 ... |
| 788 | Node 0, zone Normal 1 0 0 1 101 8 ... |
| 789 | Node 0, zone HighMem 2 0 0 1 1 0 ... |
| 790 | |
| 791 | External fragmentation is a problem under some workloads, and buddyinfo is a |
| 792 | useful tool for helping diagnose these problems. Buddyinfo will give you a |
| 793 | clue as to how big an area you can safely allocate, or why a previous |
| 794 | allocation failed. |
| 795 | |
| 796 | Each column represents the number of pages of a certain order which are |
| 797 | available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in |
| 798 | ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE |
| 799 | available in ZONE_NORMAL, etc... |
| 800 | |
| 801 | More information relevant to external fragmentation can be found in |
| 802 | pagetypeinfo. |
| 803 | |
| 804 | > cat /proc/pagetypeinfo |
| 805 | Page block order: 9 |
| 806 | Pages per block: 512 |
| 807 | |
| 808 | Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10 |
| 809 | Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0 |
| 810 | Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0 |
| 811 | Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2 |
| 812 | Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0 |
| 813 | Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0 |
| 814 | Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9 |
| 815 | Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0 |
| 816 | Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452 |
| 817 | Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0 |
| 818 | Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0 |
| 819 | |
| 820 | Number of blocks type Unmovable Reclaimable Movable Reserve Isolate |
| 821 | Node 0, zone DMA 2 0 5 1 0 |
| 822 | Node 0, zone DMA32 41 6 967 2 0 |
| 823 | |
| 824 | Fragmentation avoidance in the kernel works by grouping pages of different |
| 825 | migrate types into the same contiguous regions of memory called page blocks. |
| 826 | A page block is typically the size of the default hugepage size e.g. 2MB on |
| 827 | X86-64. By keeping pages grouped based on their ability to move, the kernel |
| 828 | can reclaim pages within a page block to satisfy a high-order allocation. |
| 829 | |
| 830 | The pagetypinfo begins with information on the size of a page block. It |
| 831 | then gives the same type of information as buddyinfo except broken down |
| 832 | by migrate-type and finishes with details on how many page blocks of each |
| 833 | type exist. |
| 834 | |
| 835 | If min_free_kbytes has been tuned correctly (recommendations made by hugeadm |
| 836 | from libhugetlbfs https://github.com/libhugetlbfs/libhugetlbfs/), one can |
| 837 | make an estimate of the likely number of huge pages that can be allocated |
| 838 | at a given point in time. All the "Movable" blocks should be allocatable |
| 839 | unless memory has been mlock()'d. Some of the Reclaimable blocks should |
| 840 | also be allocatable although a lot of filesystem metadata may have to be |
| 841 | reclaimed to achieve this. |
| 842 | |
| 843 | .............................................................................. |
| 844 | |
| 845 | meminfo: |
| 846 | |
| 847 | Provides information about distribution and utilization of memory. This |
| 848 | varies by architecture and compile options. The following is from a |
| 849 | 16GB PIII, which has highmem enabled. You may not have all of these fields. |
| 850 | |
| 851 | > cat /proc/meminfo |
| 852 | |
| 853 | MemTotal: 16344972 kB |
| 854 | MemFree: 13634064 kB |
| 855 | MemAvailable: 14836172 kB |
| 856 | Buffers: 3656 kB |
| 857 | Cached: 1195708 kB |
| 858 | SwapCached: 0 kB |
| 859 | Active: 891636 kB |
| 860 | Inactive: 1077224 kB |
| 861 | HighTotal: 15597528 kB |
| 862 | HighFree: 13629632 kB |
| 863 | LowTotal: 747444 kB |
| 864 | LowFree: 4432 kB |
| 865 | SwapTotal: 0 kB |
| 866 | SwapFree: 0 kB |
| 867 | Dirty: 968 kB |
| 868 | Writeback: 0 kB |
| 869 | AnonPages: 861800 kB |
| 870 | Mapped: 280372 kB |
| 871 | Shmem: 644 kB |
| 872 | KReclaimable: 168048 kB |
| 873 | Slab: 284364 kB |
| 874 | SReclaimable: 159856 kB |
| 875 | SUnreclaim: 124508 kB |
| 876 | PageTables: 24448 kB |
| 877 | NFS_Unstable: 0 kB |
| 878 | Bounce: 0 kB |
| 879 | WritebackTmp: 0 kB |
| 880 | CommitLimit: 7669796 kB |
| 881 | Committed_AS: 100056 kB |
| 882 | VmallocTotal: 112216 kB |
| 883 | VmallocUsed: 428 kB |
| 884 | VmallocChunk: 111088 kB |
| 885 | Percpu: 62080 kB |
| 886 | HardwareCorrupted: 0 kB |
| 887 | AnonHugePages: 49152 kB |
| 888 | ShmemHugePages: 0 kB |
| 889 | ShmemPmdMapped: 0 kB |
| 890 | |
| 891 | |
| 892 | MemTotal: Total usable ram (i.e. physical ram minus a few reserved |
| 893 | bits and the kernel binary code) |
| 894 | MemFree: The sum of LowFree+HighFree |
| 895 | MemAvailable: An estimate of how much memory is available for starting new |
| 896 | applications, without swapping. Calculated from MemFree, |
| 897 | SReclaimable, the size of the file LRU lists, and the low |
| 898 | watermarks in each zone. |
| 899 | The estimate takes into account that the system needs some |
| 900 | page cache to function well, and that not all reclaimable |
| 901 | slab will be reclaimable, due to items being in use. The |
| 902 | impact of those factors will vary from system to system. |
| 903 | Buffers: Relatively temporary storage for raw disk blocks |
| 904 | shouldn't get tremendously large (20MB or so) |
| 905 | Cached: in-memory cache for files read from the disk (the |
| 906 | pagecache). Doesn't include SwapCached |
| 907 | SwapCached: Memory that once was swapped out, is swapped back in but |
| 908 | still also is in the swapfile (if memory is needed it |
| 909 | doesn't need to be swapped out AGAIN because it is already |
| 910 | in the swapfile. This saves I/O) |
| 911 | Active: Memory that has been used more recently and usually not |
| 912 | reclaimed unless absolutely necessary. |
| 913 | Inactive: Memory which has been less recently used. It is more |
| 914 | eligible to be reclaimed for other purposes |
| 915 | HighTotal: |
| 916 | HighFree: Highmem is all memory above ~860MB of physical memory |
| 917 | Highmem areas are for use by userspace programs, or |
| 918 | for the pagecache. The kernel must use tricks to access |
| 919 | this memory, making it slower to access than lowmem. |
| 920 | LowTotal: |
| 921 | LowFree: Lowmem is memory which can be used for everything that |
| 922 | highmem can be used for, but it is also available for the |
| 923 | kernel's use for its own data structures. Among many |
| 924 | other things, it is where everything from the Slab is |
| 925 | allocated. Bad things happen when you're out of lowmem. |
| 926 | SwapTotal: total amount of swap space available |
| 927 | SwapFree: Memory which has been evicted from RAM, and is temporarily |
| 928 | on the disk |
| 929 | Dirty: Memory which is waiting to get written back to the disk |
| 930 | Writeback: Memory which is actively being written back to the disk |
| 931 | AnonPages: Non-file backed pages mapped into userspace page tables |
| 932 | HardwareCorrupted: The amount of RAM/memory in KB, the kernel identifies as |
| 933 | corrupted. |
| 934 | AnonHugePages: Non-file backed huge pages mapped into userspace page tables |
| 935 | Mapped: files which have been mmaped, such as libraries |
| 936 | Shmem: Total memory used by shared memory (shmem) and tmpfs |
| 937 | ShmemHugePages: Memory used by shared memory (shmem) and tmpfs allocated |
| 938 | with huge pages |
| 939 | ShmemPmdMapped: Shared memory mapped into userspace with huge pages |
| 940 | KReclaimable: Kernel allocations that the kernel will attempt to reclaim |
| 941 | under memory pressure. Includes SReclaimable (below), and other |
| 942 | direct allocations with a shrinker. |
| 943 | Slab: in-kernel data structures cache |
| 944 | SReclaimable: Part of Slab, that might be reclaimed, such as caches |
| 945 | SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure |
| 946 | PageTables: amount of memory dedicated to the lowest level of page |
| 947 | tables. |
| 948 | NFS_Unstable: NFS pages sent to the server, but not yet committed to stable |
| 949 | storage |
| 950 | Bounce: Memory used for block device "bounce buffers" |
| 951 | WritebackTmp: Memory used by FUSE for temporary writeback buffers |
| 952 | CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'), |
| 953 | this is the total amount of memory currently available to |
| 954 | be allocated on the system. This limit is only adhered to |
| 955 | if strict overcommit accounting is enabled (mode 2 in |
| 956 | 'vm.overcommit_memory'). |
| 957 | The CommitLimit is calculated with the following formula: |
| 958 | CommitLimit = ([total RAM pages] - [total huge TLB pages]) * |
| 959 | overcommit_ratio / 100 + [total swap pages] |
| 960 | For example, on a system with 1G of physical RAM and 7G |
| 961 | of swap with a `vm.overcommit_ratio` of 30 it would |
| 962 | yield a CommitLimit of 7.3G. |
| 963 | For more details, see the memory overcommit documentation |
| 964 | in vm/overcommit-accounting. |
| 965 | Committed_AS: The amount of memory presently allocated on the system. |
| 966 | The committed memory is a sum of all of the memory which |
| 967 | has been allocated by processes, even if it has not been |
| 968 | "used" by them as of yet. A process which malloc()'s 1G |
| 969 | of memory, but only touches 300M of it will show up as |
| 970 | using 1G. This 1G is memory which has been "committed" to |
| 971 | by the VM and can be used at any time by the allocating |
| 972 | application. With strict overcommit enabled on the system |
| 973 | (mode 2 in 'vm.overcommit_memory'),allocations which would |
| 974 | exceed the CommitLimit (detailed above) will not be permitted. |
| 975 | This is useful if one needs to guarantee that processes will |
| 976 | not fail due to lack of memory once that memory has been |
| 977 | successfully allocated. |
| 978 | VmallocTotal: total size of vmalloc memory area |
| 979 | VmallocUsed: amount of vmalloc area which is used |
| 980 | VmallocChunk: largest contiguous block of vmalloc area which is free |
| 981 | Percpu: Memory allocated to the percpu allocator used to back percpu |
| 982 | allocations. This stat excludes the cost of metadata. |
| 983 | |
| 984 | .............................................................................. |
| 985 | |
| 986 | vmallocinfo: |
| 987 | |
| 988 | Provides information about vmalloced/vmaped areas. One line per area, |
| 989 | containing the virtual address range of the area, size in bytes, |
| 990 | caller information of the creator, and optional information depending |
| 991 | on the kind of area : |
| 992 | |
| 993 | pages=nr number of pages |
| 994 | phys=addr if a physical address was specified |
| 995 | ioremap I/O mapping (ioremap() and friends) |
| 996 | vmalloc vmalloc() area |
| 997 | vmap vmap()ed pages |
| 998 | user VM_USERMAP area |
| 999 | vpages buffer for pages pointers was vmalloced (huge area) |
| 1000 | N<node>=nr (Only on NUMA kernels) |
| 1001 | Number of pages allocated on memory node <node> |
| 1002 | |
| 1003 | > cat /proc/vmallocinfo |
| 1004 | 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ... |
| 1005 | /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128 |
| 1006 | 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ... |
| 1007 | /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64 |
| 1008 | 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f... |
| 1009 | phys=7fee8000 ioremap |
| 1010 | 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f... |
| 1011 | phys=7fee7000 ioremap |
| 1012 | 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210 |
| 1013 | 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ... |
| 1014 | /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3 |
| 1015 | 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ... |
| 1016 | pages=2 vmalloc N1=2 |
| 1017 | 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ... |
| 1018 | /0x130 [x_tables] pages=4 vmalloc N0=4 |
| 1019 | 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ... |
| 1020 | pages=14 vmalloc N2=14 |
| 1021 | 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ... |
| 1022 | pages=4 vmalloc N1=4 |
| 1023 | 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ... |
| 1024 | pages=2 vmalloc N1=2 |
| 1025 | 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ... |
| 1026 | pages=10 vmalloc N0=10 |
| 1027 | |
| 1028 | .............................................................................. |
| 1029 | |
| 1030 | softirqs: |
| 1031 | |
| 1032 | Provides counts of softirq handlers serviced since boot time, for each cpu. |
| 1033 | |
| 1034 | > cat /proc/softirqs |
| 1035 | CPU0 CPU1 CPU2 CPU3 |
| 1036 | HI: 0 0 0 0 |
| 1037 | TIMER: 27166 27120 27097 27034 |
| 1038 | NET_TX: 0 0 0 17 |
| 1039 | NET_RX: 42 0 0 39 |
| 1040 | BLOCK: 0 0 107 1121 |
| 1041 | TASKLET: 0 0 0 290 |
| 1042 | SCHED: 27035 26983 26971 26746 |
| 1043 | HRTIMER: 0 0 0 0 |
| 1044 | RCU: 1678 1769 2178 2250 |
| 1045 | |
| 1046 | |
| 1047 | 1.3 IDE devices in /proc/ide |
| 1048 | ---------------------------- |
| 1049 | |
| 1050 | The subdirectory /proc/ide contains information about all IDE devices of which |
| 1051 | the kernel is aware. There is one subdirectory for each IDE controller, the |
| 1052 | file drivers and a link for each IDE device, pointing to the device directory |
| 1053 | in the controller specific subtree. |
| 1054 | |
| 1055 | The file drivers contains general information about the drivers used for the |
| 1056 | IDE devices: |
| 1057 | |
| 1058 | > cat /proc/ide/drivers |
| 1059 | ide-cdrom version 4.53 |
| 1060 | ide-disk version 1.08 |
| 1061 | |
| 1062 | More detailed information can be found in the controller specific |
| 1063 | subdirectories. These are named ide0, ide1 and so on. Each of these |
| 1064 | directories contains the files shown in table 1-6. |
| 1065 | |
| 1066 | |
| 1067 | Table 1-6: IDE controller info in /proc/ide/ide? |
| 1068 | .............................................................................. |
| 1069 | File Content |
| 1070 | channel IDE channel (0 or 1) |
| 1071 | config Configuration (only for PCI/IDE bridge) |
| 1072 | mate Mate name |
| 1073 | model Type/Chipset of IDE controller |
| 1074 | .............................................................................. |
| 1075 | |
| 1076 | Each device connected to a controller has a separate subdirectory in the |
| 1077 | controllers directory. The files listed in table 1-7 are contained in these |
| 1078 | directories. |
| 1079 | |
| 1080 | |
| 1081 | Table 1-7: IDE device information |
| 1082 | .............................................................................. |
| 1083 | File Content |
| 1084 | cache The cache |
| 1085 | capacity Capacity of the medium (in 512Byte blocks) |
| 1086 | driver driver and version |
| 1087 | geometry physical and logical geometry |
| 1088 | identify device identify block |
| 1089 | media media type |
| 1090 | model device identifier |
| 1091 | settings device setup |
| 1092 | smart_thresholds IDE disk management thresholds |
| 1093 | smart_values IDE disk management values |
| 1094 | .............................................................................. |
| 1095 | |
| 1096 | The most interesting file is settings. This file contains a nice overview of |
| 1097 | the drive parameters: |
| 1098 | |
| 1099 | # cat /proc/ide/ide0/hda/settings |
| 1100 | name value min max mode |
| 1101 | ---- ----- --- --- ---- |
| 1102 | bios_cyl 526 0 65535 rw |
| 1103 | bios_head 255 0 255 rw |
| 1104 | bios_sect 63 0 63 rw |
| 1105 | breada_readahead 4 0 127 rw |
| 1106 | bswap 0 0 1 r |
| 1107 | file_readahead 72 0 2097151 rw |
| 1108 | io_32bit 0 0 3 rw |
| 1109 | keepsettings 0 0 1 rw |
| 1110 | max_kb_per_request 122 1 127 rw |
| 1111 | multcount 0 0 8 rw |
| 1112 | nice1 1 0 1 rw |
| 1113 | nowerr 0 0 1 rw |
| 1114 | pio_mode write-only 0 255 w |
| 1115 | slow 0 0 1 rw |
| 1116 | unmaskirq 0 0 1 rw |
| 1117 | using_dma 0 0 1 rw |
| 1118 | |
| 1119 | |
| 1120 | 1.4 Networking info in /proc/net |
| 1121 | -------------------------------- |
| 1122 | |
| 1123 | The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the |
| 1124 | additional values you get for IP version 6 if you configure the kernel to |
| 1125 | support this. Table 1-9 lists the files and their meaning. |
| 1126 | |
| 1127 | |
| 1128 | Table 1-8: IPv6 info in /proc/net |
| 1129 | .............................................................................. |
| 1130 | File Content |
| 1131 | udp6 UDP sockets (IPv6) |
| 1132 | tcp6 TCP sockets (IPv6) |
| 1133 | raw6 Raw device statistics (IPv6) |
| 1134 | igmp6 IP multicast addresses, which this host joined (IPv6) |
| 1135 | if_inet6 List of IPv6 interface addresses |
| 1136 | ipv6_route Kernel routing table for IPv6 |
| 1137 | rt6_stats Global IPv6 routing tables statistics |
| 1138 | sockstat6 Socket statistics (IPv6) |
| 1139 | snmp6 Snmp data (IPv6) |
| 1140 | .............................................................................. |
| 1141 | |
| 1142 | |
| 1143 | Table 1-9: Network info in /proc/net |
| 1144 | .............................................................................. |
| 1145 | File Content |
| 1146 | arp Kernel ARP table |
| 1147 | dev network devices with statistics |
| 1148 | dev_mcast the Layer2 multicast groups a device is listening too |
| 1149 | (interface index, label, number of references, number of bound |
| 1150 | addresses). |
| 1151 | dev_stat network device status |
| 1152 | ip_fwchains Firewall chain linkage |
| 1153 | ip_fwnames Firewall chain names |
| 1154 | ip_masq Directory containing the masquerading tables |
| 1155 | ip_masquerade Major masquerading table |
| 1156 | netstat Network statistics |
| 1157 | raw raw device statistics |
| 1158 | route Kernel routing table |
| 1159 | rpc Directory containing rpc info |
| 1160 | rt_cache Routing cache |
| 1161 | snmp SNMP data |
| 1162 | sockstat Socket statistics |
| 1163 | tcp TCP sockets |
| 1164 | udp UDP sockets |
| 1165 | unix UNIX domain sockets |
| 1166 | wireless Wireless interface data (Wavelan etc) |
| 1167 | igmp IP multicast addresses, which this host joined |
| 1168 | psched Global packet scheduler parameters. |
| 1169 | netlink List of PF_NETLINK sockets |
| 1170 | ip_mr_vifs List of multicast virtual interfaces |
| 1171 | ip_mr_cache List of multicast routing cache |
| 1172 | .............................................................................. |
| 1173 | |
| 1174 | You can use this information to see which network devices are available in |
| 1175 | your system and how much traffic was routed over those devices: |
| 1176 | |
| 1177 | > cat /proc/net/dev |
| 1178 | Inter-|Receive |[... |
| 1179 | face |bytes packets errs drop fifo frame compressed multicast|[... |
| 1180 | lo: 908188 5596 0 0 0 0 0 0 [... |
| 1181 | ppp0:15475140 20721 410 0 0 410 0 0 [... |
| 1182 | eth0: 614530 7085 0 0 0 0 0 1 [... |
| 1183 | |
| 1184 | ...] Transmit |
| 1185 | ...] bytes packets errs drop fifo colls carrier compressed |
| 1186 | ...] 908188 5596 0 0 0 0 0 0 |
| 1187 | ...] 1375103 17405 0 0 0 0 0 0 |
| 1188 | ...] 1703981 5535 0 0 0 3 0 0 |
| 1189 | |
| 1190 | In addition, each Channel Bond interface has its own directory. For |
| 1191 | example, the bond0 device will have a directory called /proc/net/bond0/. |
| 1192 | It will contain information that is specific to that bond, such as the |
| 1193 | current slaves of the bond, the link status of the slaves, and how |
| 1194 | many times the slaves link has failed. |
| 1195 | |
| 1196 | 1.5 SCSI info |
| 1197 | ------------- |
| 1198 | |
| 1199 | If you have a SCSI host adapter in your system, you'll find a subdirectory |
| 1200 | named after the driver for this adapter in /proc/scsi. You'll also see a list |
| 1201 | of all recognized SCSI devices in /proc/scsi: |
| 1202 | |
| 1203 | >cat /proc/scsi/scsi |
| 1204 | Attached devices: |
| 1205 | Host: scsi0 Channel: 00 Id: 00 Lun: 00 |
| 1206 | Vendor: IBM Model: DGHS09U Rev: 03E0 |
| 1207 | Type: Direct-Access ANSI SCSI revision: 03 |
| 1208 | Host: scsi0 Channel: 00 Id: 06 Lun: 00 |
| 1209 | Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04 |
| 1210 | Type: CD-ROM ANSI SCSI revision: 02 |
| 1211 | |
| 1212 | |
| 1213 | The directory named after the driver has one file for each adapter found in |
| 1214 | the system. These files contain information about the controller, including |
| 1215 | the used IRQ and the IO address range. The amount of information shown is |
| 1216 | dependent on the adapter you use. The example shows the output for an Adaptec |
| 1217 | AHA-2940 SCSI adapter: |
| 1218 | |
| 1219 | > cat /proc/scsi/aic7xxx/0 |
| 1220 | |
| 1221 | Adaptec AIC7xxx driver version: 5.1.19/3.2.4 |
| 1222 | Compile Options: |
| 1223 | TCQ Enabled By Default : Disabled |
| 1224 | AIC7XXX_PROC_STATS : Disabled |
| 1225 | AIC7XXX_RESET_DELAY : 5 |
| 1226 | Adapter Configuration: |
| 1227 | SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter |
| 1228 | Ultra Wide Controller |
| 1229 | PCI MMAPed I/O Base: 0xeb001000 |
| 1230 | Adapter SEEPROM Config: SEEPROM found and used. |
| 1231 | Adaptec SCSI BIOS: Enabled |
| 1232 | IRQ: 10 |
| 1233 | SCBs: Active 0, Max Active 2, |
| 1234 | Allocated 15, HW 16, Page 255 |
| 1235 | Interrupts: 160328 |
| 1236 | BIOS Control Word: 0x18b6 |
| 1237 | Adapter Control Word: 0x005b |
| 1238 | Extended Translation: Enabled |
| 1239 | Disconnect Enable Flags: 0xffff |
| 1240 | Ultra Enable Flags: 0x0001 |
| 1241 | Tag Queue Enable Flags: 0x0000 |
| 1242 | Ordered Queue Tag Flags: 0x0000 |
| 1243 | Default Tag Queue Depth: 8 |
| 1244 | Tagged Queue By Device array for aic7xxx host instance 0: |
| 1245 | {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255} |
| 1246 | Actual queue depth per device for aic7xxx host instance 0: |
| 1247 | {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} |
| 1248 | Statistics: |
| 1249 | (scsi0:0:0:0) |
| 1250 | Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8 |
| 1251 | Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0) |
| 1252 | Total transfers 160151 (74577 reads and 85574 writes) |
| 1253 | (scsi0:0:6:0) |
| 1254 | Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15 |
| 1255 | Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0) |
| 1256 | Total transfers 0 (0 reads and 0 writes) |
| 1257 | |
| 1258 | |
| 1259 | 1.6 Parallel port info in /proc/parport |
| 1260 | --------------------------------------- |
| 1261 | |
| 1262 | The directory /proc/parport contains information about the parallel ports of |
| 1263 | your system. It has one subdirectory for each port, named after the port |
| 1264 | number (0,1,2,...). |
| 1265 | |
| 1266 | These directories contain the four files shown in Table 1-10. |
| 1267 | |
| 1268 | |
| 1269 | Table 1-10: Files in /proc/parport |
| 1270 | .............................................................................. |
| 1271 | File Content |
| 1272 | autoprobe Any IEEE-1284 device ID information that has been acquired. |
| 1273 | devices list of the device drivers using that port. A + will appear by the |
| 1274 | name of the device currently using the port (it might not appear |
| 1275 | against any). |
| 1276 | hardware Parallel port's base address, IRQ line and DMA channel. |
| 1277 | irq IRQ that parport is using for that port. This is in a separate |
| 1278 | file to allow you to alter it by writing a new value in (IRQ |
| 1279 | number or none). |
| 1280 | .............................................................................. |
| 1281 | |
| 1282 | 1.7 TTY info in /proc/tty |
| 1283 | ------------------------- |
| 1284 | |
| 1285 | Information about the available and actually used tty's can be found in the |
| 1286 | directory /proc/tty.You'll find entries for drivers and line disciplines in |
| 1287 | this directory, as shown in Table 1-11. |
| 1288 | |
| 1289 | |
| 1290 | Table 1-11: Files in /proc/tty |
| 1291 | .............................................................................. |
| 1292 | File Content |
| 1293 | drivers list of drivers and their usage |
| 1294 | ldiscs registered line disciplines |
| 1295 | driver/serial usage statistic and status of single tty lines |
| 1296 | .............................................................................. |
| 1297 | |
| 1298 | To see which tty's are currently in use, you can simply look into the file |
| 1299 | /proc/tty/drivers: |
| 1300 | |
| 1301 | > cat /proc/tty/drivers |
| 1302 | pty_slave /dev/pts 136 0-255 pty:slave |
| 1303 | pty_master /dev/ptm 128 0-255 pty:master |
| 1304 | pty_slave /dev/ttyp 3 0-255 pty:slave |
| 1305 | pty_master /dev/pty 2 0-255 pty:master |
| 1306 | serial /dev/cua 5 64-67 serial:callout |
| 1307 | serial /dev/ttyS 4 64-67 serial |
| 1308 | /dev/tty0 /dev/tty0 4 0 system:vtmaster |
| 1309 | /dev/ptmx /dev/ptmx 5 2 system |
| 1310 | /dev/console /dev/console 5 1 system:console |
| 1311 | /dev/tty /dev/tty 5 0 system:/dev/tty |
| 1312 | unknown /dev/tty 4 1-63 console |
| 1313 | |
| 1314 | |
| 1315 | 1.8 Miscellaneous kernel statistics in /proc/stat |
| 1316 | ------------------------------------------------- |
| 1317 | |
| 1318 | Various pieces of information about kernel activity are available in the |
| 1319 | /proc/stat file. All of the numbers reported in this file are aggregates |
| 1320 | since the system first booted. For a quick look, simply cat the file: |
| 1321 | |
| 1322 | > cat /proc/stat |
| 1323 | cpu 2255 34 2290 22625563 6290 127 456 0 0 0 |
| 1324 | cpu0 1132 34 1441 11311718 3675 127 438 0 0 0 |
| 1325 | cpu1 1123 0 849 11313845 2614 0 18 0 0 0 |
| 1326 | intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...] |
| 1327 | ctxt 1990473 |
| 1328 | btime 1062191376 |
| 1329 | processes 2915 |
| 1330 | procs_running 1 |
| 1331 | procs_blocked 0 |
| 1332 | softirq 183433 0 21755 12 39 1137 231 21459 2263 |
| 1333 | |
| 1334 | The very first "cpu" line aggregates the numbers in all of the other "cpuN" |
| 1335 | lines. These numbers identify the amount of time the CPU has spent performing |
| 1336 | different kinds of work. Time units are in USER_HZ (typically hundredths of a |
| 1337 | second). The meanings of the columns are as follows, from left to right: |
| 1338 | |
| 1339 | - user: normal processes executing in user mode |
| 1340 | - nice: niced processes executing in user mode |
| 1341 | - system: processes executing in kernel mode |
| 1342 | - idle: twiddling thumbs |
| 1343 | - iowait: In a word, iowait stands for waiting for I/O to complete. But there |
| 1344 | are several problems: |
| 1345 | 1. Cpu will not wait for I/O to complete, iowait is the time that a task is |
| 1346 | waiting for I/O to complete. When cpu goes into idle state for |
| 1347 | outstanding task io, another task will be scheduled on this CPU. |
| 1348 | 2. In a multi-core CPU, the task waiting for I/O to complete is not running |
| 1349 | on any CPU, so the iowait of each CPU is difficult to calculate. |
| 1350 | 3. The value of iowait field in /proc/stat will decrease in certain |
| 1351 | conditions. |
| 1352 | So, the iowait is not reliable by reading from /proc/stat. |
| 1353 | - irq: servicing interrupts |
| 1354 | - softirq: servicing softirqs |
| 1355 | - steal: involuntary wait |
| 1356 | - guest: running a normal guest |
| 1357 | - guest_nice: running a niced guest |
| 1358 | |
| 1359 | The "intr" line gives counts of interrupts serviced since boot time, for each |
| 1360 | of the possible system interrupts. The first column is the total of all |
| 1361 | interrupts serviced including unnumbered architecture specific interrupts; |
| 1362 | each subsequent column is the total for that particular numbered interrupt. |
| 1363 | Unnumbered interrupts are not shown, only summed into the total. |
| 1364 | |
| 1365 | The "ctxt" line gives the total number of context switches across all CPUs. |
| 1366 | |
| 1367 | The "btime" line gives the time at which the system booted, in seconds since |
| 1368 | the Unix epoch. |
| 1369 | |
| 1370 | The "processes" line gives the number of processes and threads created, which |
| 1371 | includes (but is not limited to) those created by calls to the fork() and |
| 1372 | clone() system calls. |
| 1373 | |
| 1374 | The "procs_running" line gives the total number of threads that are |
| 1375 | running or ready to run (i.e., the total number of runnable threads). |
| 1376 | |
| 1377 | The "procs_blocked" line gives the number of processes currently blocked, |
| 1378 | waiting for I/O to complete. |
| 1379 | |
| 1380 | The "softirq" line gives counts of softirqs serviced since boot time, for each |
| 1381 | of the possible system softirqs. The first column is the total of all |
| 1382 | softirqs serviced; each subsequent column is the total for that particular |
| 1383 | softirq. |
| 1384 | |
| 1385 | |
| 1386 | 1.9 Ext4 file system parameters |
| 1387 | ------------------------------- |
| 1388 | |
| 1389 | Information about mounted ext4 file systems can be found in |
| 1390 | /proc/fs/ext4. Each mounted filesystem will have a directory in |
| 1391 | /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or |
| 1392 | /proc/fs/ext4/dm-0). The files in each per-device directory are shown |
| 1393 | in Table 1-12, below. |
| 1394 | |
| 1395 | Table 1-12: Files in /proc/fs/ext4/<devname> |
| 1396 | .............................................................................. |
| 1397 | File Content |
| 1398 | mb_groups details of multiblock allocator buddy cache of free blocks |
| 1399 | .............................................................................. |
| 1400 | |
| 1401 | 2.0 /proc/consoles |
| 1402 | ------------------ |
| 1403 | Shows registered system console lines. |
| 1404 | |
| 1405 | To see which character device lines are currently used for the system console |
| 1406 | /dev/console, you may simply look into the file /proc/consoles: |
| 1407 | |
| 1408 | > cat /proc/consoles |
| 1409 | tty0 -WU (ECp) 4:7 |
| 1410 | ttyS0 -W- (Ep) 4:64 |
| 1411 | |
| 1412 | The columns are: |
| 1413 | |
| 1414 | device name of the device |
| 1415 | operations R = can do read operations |
| 1416 | W = can do write operations |
| 1417 | U = can do unblank |
| 1418 | flags E = it is enabled |
| 1419 | C = it is preferred console |
| 1420 | B = it is primary boot console |
| 1421 | p = it is used for printk buffer |
| 1422 | b = it is not a TTY but a Braille device |
| 1423 | a = it is safe to use when cpu is offline |
| 1424 | major:minor major and minor number of the device separated by a colon |
| 1425 | |
| 1426 | ------------------------------------------------------------------------------ |
| 1427 | Summary |
| 1428 | ------------------------------------------------------------------------------ |
| 1429 | The /proc file system serves information about the running system. It not only |
| 1430 | allows access to process data but also allows you to request the kernel status |
| 1431 | by reading files in the hierarchy. |
| 1432 | |
| 1433 | The directory structure of /proc reflects the types of information and makes |
| 1434 | it easy, if not obvious, where to look for specific data. |
| 1435 | ------------------------------------------------------------------------------ |
| 1436 | |
| 1437 | ------------------------------------------------------------------------------ |
| 1438 | CHAPTER 2: MODIFYING SYSTEM PARAMETERS |
| 1439 | ------------------------------------------------------------------------------ |
| 1440 | |
| 1441 | ------------------------------------------------------------------------------ |
| 1442 | In This Chapter |
| 1443 | ------------------------------------------------------------------------------ |
| 1444 | * Modifying kernel parameters by writing into files found in /proc/sys |
| 1445 | * Exploring the files which modify certain parameters |
| 1446 | * Review of the /proc/sys file tree |
| 1447 | ------------------------------------------------------------------------------ |
| 1448 | |
| 1449 | |
| 1450 | A very interesting part of /proc is the directory /proc/sys. This is not only |
| 1451 | a source of information, it also allows you to change parameters within the |
| 1452 | kernel. Be very careful when attempting this. You can optimize your system, |
| 1453 | but you can also cause it to crash. Never alter kernel parameters on a |
| 1454 | production system. Set up a development machine and test to make sure that |
| 1455 | everything works the way you want it to. You may have no alternative but to |
| 1456 | reboot the machine once an error has been made. |
| 1457 | |
| 1458 | To change a value, simply echo the new value into the file. An example is |
| 1459 | given below in the section on the file system data. You need to be root to do |
| 1460 | this. You can create your own boot script to perform this every time your |
| 1461 | system boots. |
| 1462 | |
| 1463 | The files in /proc/sys can be used to fine tune and monitor miscellaneous and |
| 1464 | general things in the operation of the Linux kernel. Since some of the files |
| 1465 | can inadvertently disrupt your system, it is advisable to read both |
| 1466 | documentation and source before actually making adjustments. In any case, be |
| 1467 | very careful when writing to any of these files. The entries in /proc may |
| 1468 | change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt |
| 1469 | review the kernel documentation in the directory /usr/src/linux/Documentation. |
| 1470 | This chapter is heavily based on the documentation included in the pre 2.2 |
| 1471 | kernels, and became part of it in version 2.2.1 of the Linux kernel. |
| 1472 | |
| 1473 | Please see: Documentation/sysctl/ directory for descriptions of these |
| 1474 | entries. |
| 1475 | |
| 1476 | ------------------------------------------------------------------------------ |
| 1477 | Summary |
| 1478 | ------------------------------------------------------------------------------ |
| 1479 | Certain aspects of kernel behavior can be modified at runtime, without the |
| 1480 | need to recompile the kernel, or even to reboot the system. The files in the |
| 1481 | /proc/sys tree can not only be read, but also modified. You can use the echo |
| 1482 | command to write value into these files, thereby changing the default settings |
| 1483 | of the kernel. |
| 1484 | ------------------------------------------------------------------------------ |
| 1485 | |
| 1486 | ------------------------------------------------------------------------------ |
| 1487 | CHAPTER 3: PER-PROCESS PARAMETERS |
| 1488 | ------------------------------------------------------------------------------ |
| 1489 | |
| 1490 | 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score |
| 1491 | -------------------------------------------------------------------------------- |
| 1492 | |
| 1493 | These file can be used to adjust the badness heuristic used to select which |
| 1494 | process gets killed in out of memory conditions. |
| 1495 | |
| 1496 | The badness heuristic assigns a value to each candidate task ranging from 0 |
| 1497 | (never kill) to 1000 (always kill) to determine which process is targeted. The |
| 1498 | units are roughly a proportion along that range of allowed memory the process |
| 1499 | may allocate from based on an estimation of its current memory and swap use. |
| 1500 | For example, if a task is using all allowed memory, its badness score will be |
| 1501 | 1000. If it is using half of its allowed memory, its score will be 500. |
| 1502 | |
| 1503 | There is an additional factor included in the badness score: the current memory |
| 1504 | and swap usage is discounted by 3% for root processes. |
| 1505 | |
| 1506 | The amount of "allowed" memory depends on the context in which the oom killer |
| 1507 | was called. If it is due to the memory assigned to the allocating task's cpuset |
| 1508 | being exhausted, the allowed memory represents the set of mems assigned to that |
| 1509 | cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed |
| 1510 | memory represents the set of mempolicy nodes. If it is due to a memory |
| 1511 | limit (or swap limit) being reached, the allowed memory is that configured |
| 1512 | limit. Finally, if it is due to the entire system being out of memory, the |
| 1513 | allowed memory represents all allocatable resources. |
| 1514 | |
| 1515 | The value of /proc/<pid>/oom_score_adj is added to the badness score before it |
| 1516 | is used to determine which task to kill. Acceptable values range from -1000 |
| 1517 | (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to |
| 1518 | polarize the preference for oom killing either by always preferring a certain |
| 1519 | task or completely disabling it. The lowest possible value, -1000, is |
| 1520 | equivalent to disabling oom killing entirely for that task since it will always |
| 1521 | report a badness score of 0. |
| 1522 | |
| 1523 | Consequently, it is very simple for userspace to define the amount of memory to |
| 1524 | consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for |
| 1525 | example, is roughly equivalent to allowing the remainder of tasks sharing the |
| 1526 | same system, cpuset, mempolicy, or memory controller resources to use at least |
| 1527 | 50% more memory. A value of -500, on the other hand, would be roughly |
| 1528 | equivalent to discounting 50% of the task's allowed memory from being considered |
| 1529 | as scoring against the task. |
| 1530 | |
| 1531 | For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also |
| 1532 | be used to tune the badness score. Its acceptable values range from -16 |
| 1533 | (OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17 |
| 1534 | (OOM_DISABLE) to disable oom killing entirely for that task. Its value is |
| 1535 | scaled linearly with /proc/<pid>/oom_score_adj. |
| 1536 | |
| 1537 | The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last |
| 1538 | value set by a CAP_SYS_RESOURCE process. To reduce the value any lower |
| 1539 | requires CAP_SYS_RESOURCE. |
| 1540 | |
| 1541 | Caveat: when a parent task is selected, the oom killer will sacrifice any first |
| 1542 | generation children with separate address spaces instead, if possible. This |
| 1543 | avoids servers and important system daemons from being killed and loses the |
| 1544 | minimal amount of work. |
| 1545 | |
| 1546 | |
| 1547 | 3.2 /proc/<pid>/oom_score - Display current oom-killer score |
| 1548 | ------------------------------------------------------------- |
| 1549 | |
| 1550 | This file can be used to check the current score used by the oom-killer is for |
| 1551 | any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which |
| 1552 | process should be killed in an out-of-memory situation. |
| 1553 | |
| 1554 | |
| 1555 | 3.3 /proc/<pid>/io - Display the IO accounting fields |
| 1556 | ------------------------------------------------------- |
| 1557 | |
| 1558 | This file contains IO statistics for each running process |
| 1559 | |
| 1560 | Example |
| 1561 | ------- |
| 1562 | |
| 1563 | test:/tmp # dd if=/dev/zero of=/tmp/test.dat & |
| 1564 | [1] 3828 |
| 1565 | |
| 1566 | test:/tmp # cat /proc/3828/io |
| 1567 | rchar: 323934931 |
| 1568 | wchar: 323929600 |
| 1569 | syscr: 632687 |
| 1570 | syscw: 632675 |
| 1571 | read_bytes: 0 |
| 1572 | write_bytes: 323932160 |
| 1573 | cancelled_write_bytes: 0 |
| 1574 | |
| 1575 | |
| 1576 | Description |
| 1577 | ----------- |
| 1578 | |
| 1579 | rchar |
| 1580 | ----- |
| 1581 | |
| 1582 | I/O counter: chars read |
| 1583 | The number of bytes which this task has caused to be read from storage. This |
| 1584 | is simply the sum of bytes which this process passed to read() and pread(). |
| 1585 | It includes things like tty IO and it is unaffected by whether or not actual |
| 1586 | physical disk IO was required (the read might have been satisfied from |
| 1587 | pagecache) |
| 1588 | |
| 1589 | |
| 1590 | wchar |
| 1591 | ----- |
| 1592 | |
| 1593 | I/O counter: chars written |
| 1594 | The number of bytes which this task has caused, or shall cause to be written |
| 1595 | to disk. Similar caveats apply here as with rchar. |
| 1596 | |
| 1597 | |
| 1598 | syscr |
| 1599 | ----- |
| 1600 | |
| 1601 | I/O counter: read syscalls |
| 1602 | Attempt to count the number of read I/O operations, i.e. syscalls like read() |
| 1603 | and pread(). |
| 1604 | |
| 1605 | |
| 1606 | syscw |
| 1607 | ----- |
| 1608 | |
| 1609 | I/O counter: write syscalls |
| 1610 | Attempt to count the number of write I/O operations, i.e. syscalls like |
| 1611 | write() and pwrite(). |
| 1612 | |
| 1613 | |
| 1614 | read_bytes |
| 1615 | ---------- |
| 1616 | |
| 1617 | I/O counter: bytes read |
| 1618 | Attempt to count the number of bytes which this process really did cause to |
| 1619 | be fetched from the storage layer. Done at the submit_bio() level, so it is |
| 1620 | accurate for block-backed filesystems. <please add status regarding NFS and |
| 1621 | CIFS at a later time> |
| 1622 | |
| 1623 | |
| 1624 | write_bytes |
| 1625 | ----------- |
| 1626 | |
| 1627 | I/O counter: bytes written |
| 1628 | Attempt to count the number of bytes which this process caused to be sent to |
| 1629 | the storage layer. This is done at page-dirtying time. |
| 1630 | |
| 1631 | |
| 1632 | cancelled_write_bytes |
| 1633 | --------------------- |
| 1634 | |
| 1635 | The big inaccuracy here is truncate. If a process writes 1MB to a file and |
| 1636 | then deletes the file, it will in fact perform no writeout. But it will have |
| 1637 | been accounted as having caused 1MB of write. |
| 1638 | In other words: The number of bytes which this process caused to not happen, |
| 1639 | by truncating pagecache. A task can cause "negative" IO too. If this task |
| 1640 | truncates some dirty pagecache, some IO which another task has been accounted |
| 1641 | for (in its write_bytes) will not be happening. We _could_ just subtract that |
| 1642 | from the truncating task's write_bytes, but there is information loss in doing |
| 1643 | that. |
| 1644 | |
| 1645 | |
| 1646 | Note |
| 1647 | ---- |
| 1648 | |
| 1649 | At its current implementation state, this is a bit racy on 32-bit machines: if |
| 1650 | process A reads process B's /proc/pid/io while process B is updating one of |
| 1651 | those 64-bit counters, process A could see an intermediate result. |
| 1652 | |
| 1653 | |
| 1654 | More information about this can be found within the taskstats documentation in |
| 1655 | Documentation/accounting. |
| 1656 | |
| 1657 | 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings |
| 1658 | --------------------------------------------------------------- |
| 1659 | When a process is dumped, all anonymous memory is written to a core file as |
| 1660 | long as the size of the core file isn't limited. But sometimes we don't want |
| 1661 | to dump some memory segments, for example, huge shared memory or DAX. |
| 1662 | Conversely, sometimes we want to save file-backed memory segments into a core |
| 1663 | file, not only the individual files. |
| 1664 | |
| 1665 | /proc/<pid>/coredump_filter allows you to customize which memory segments |
| 1666 | will be dumped when the <pid> process is dumped. coredump_filter is a bitmask |
| 1667 | of memory types. If a bit of the bitmask is set, memory segments of the |
| 1668 | corresponding memory type are dumped, otherwise they are not dumped. |
| 1669 | |
| 1670 | The following 9 memory types are supported: |
| 1671 | - (bit 0) anonymous private memory |
| 1672 | - (bit 1) anonymous shared memory |
| 1673 | - (bit 2) file-backed private memory |
| 1674 | - (bit 3) file-backed shared memory |
| 1675 | - (bit 4) ELF header pages in file-backed private memory areas (it is |
| 1676 | effective only if the bit 2 is cleared) |
| 1677 | - (bit 5) hugetlb private memory |
| 1678 | - (bit 6) hugetlb shared memory |
| 1679 | - (bit 7) DAX private memory |
| 1680 | - (bit 8) DAX shared memory |
| 1681 | |
| 1682 | Note that MMIO pages such as frame buffer are never dumped and vDSO pages |
| 1683 | are always dumped regardless of the bitmask status. |
| 1684 | |
| 1685 | Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is |
| 1686 | only affected by bit 5-6, and DAX is only affected by bits 7-8. |
| 1687 | |
| 1688 | The default value of coredump_filter is 0x33; this means all anonymous memory |
| 1689 | segments, ELF header pages and hugetlb private memory are dumped. |
| 1690 | |
| 1691 | If you don't want to dump all shared memory segments attached to pid 1234, |
| 1692 | write 0x31 to the process's proc file. |
| 1693 | |
| 1694 | $ echo 0x31 > /proc/1234/coredump_filter |
| 1695 | |
| 1696 | When a new process is created, the process inherits the bitmask status from its |
| 1697 | parent. It is useful to set up coredump_filter before the program runs. |
| 1698 | For example: |
| 1699 | |
| 1700 | $ echo 0x7 > /proc/self/coredump_filter |
| 1701 | $ ./some_program |
| 1702 | |
| 1703 | 3.5 /proc/<pid>/mountinfo - Information about mounts |
| 1704 | -------------------------------------------------------- |
| 1705 | |
| 1706 | This file contains lines of the form: |
| 1707 | |
| 1708 | 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue |
| 1709 | (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11) |
| 1710 | |
| 1711 | (1) mount ID: unique identifier of the mount (may be reused after umount) |
| 1712 | (2) parent ID: ID of parent (or of self for the top of the mount tree) |
| 1713 | (3) major:minor: value of st_dev for files on filesystem |
| 1714 | (4) root: root of the mount within the filesystem |
| 1715 | (5) mount point: mount point relative to the process's root |
| 1716 | (6) mount options: per mount options |
| 1717 | (7) optional fields: zero or more fields of the form "tag[:value]" |
| 1718 | (8) separator: marks the end of the optional fields |
| 1719 | (9) filesystem type: name of filesystem of the form "type[.subtype]" |
| 1720 | (10) mount source: filesystem specific information or "none" |
| 1721 | (11) super options: per super block options |
| 1722 | |
| 1723 | Parsers should ignore all unrecognised optional fields. Currently the |
| 1724 | possible optional fields are: |
| 1725 | |
| 1726 | shared:X mount is shared in peer group X |
| 1727 | master:X mount is slave to peer group X |
| 1728 | propagate_from:X mount is slave and receives propagation from peer group X (*) |
| 1729 | unbindable mount is unbindable |
| 1730 | |
| 1731 | (*) X is the closest dominant peer group under the process's root. If |
| 1732 | X is the immediate master of the mount, or if there's no dominant peer |
| 1733 | group under the same root, then only the "master:X" field is present |
| 1734 | and not the "propagate_from:X" field. |
| 1735 | |
| 1736 | For more information on mount propagation see: |
| 1737 | |
| 1738 | Documentation/filesystems/sharedsubtree.txt |
| 1739 | |
| 1740 | |
| 1741 | 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm |
| 1742 | -------------------------------------------------------- |
| 1743 | These files provide a method to access a tasks comm value. It also allows for |
| 1744 | a task to set its own or one of its thread siblings comm value. The comm value |
| 1745 | is limited in size compared to the cmdline value, so writing anything longer |
| 1746 | then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated |
| 1747 | comm value. |
| 1748 | |
| 1749 | |
| 1750 | 3.7 /proc/<pid>/task/<tid>/children - Information about task children |
| 1751 | ------------------------------------------------------------------------- |
| 1752 | This file provides a fast way to retrieve first level children pids |
| 1753 | of a task pointed by <pid>/<tid> pair. The format is a space separated |
| 1754 | stream of pids. |
| 1755 | |
| 1756 | Note the "first level" here -- if a child has own children they will |
| 1757 | not be listed here, one needs to read /proc/<children-pid>/task/<tid>/children |
| 1758 | to obtain the descendants. |
| 1759 | |
| 1760 | Since this interface is intended to be fast and cheap it doesn't |
| 1761 | guarantee to provide precise results and some children might be |
| 1762 | skipped, especially if they've exited right after we printed their |
| 1763 | pids, so one need to either stop or freeze processes being inspected |
| 1764 | if precise results are needed. |
| 1765 | |
| 1766 | |
| 1767 | 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file |
| 1768 | --------------------------------------------------------------- |
| 1769 | This file provides information associated with an opened file. The regular |
| 1770 | files have at least three fields -- 'pos', 'flags' and mnt_id. The 'pos' |
| 1771 | represents the current offset of the opened file in decimal form [see lseek(2) |
| 1772 | for details], 'flags' denotes the octal O_xxx mask the file has been |
| 1773 | created with [see open(2) for details] and 'mnt_id' represents mount ID of |
| 1774 | the file system containing the opened file [see 3.5 /proc/<pid>/mountinfo |
| 1775 | for details]. |
| 1776 | |
| 1777 | A typical output is |
| 1778 | |
| 1779 | pos: 0 |
| 1780 | flags: 0100002 |
| 1781 | mnt_id: 19 |
| 1782 | |
| 1783 | All locks associated with a file descriptor are shown in its fdinfo too. |
| 1784 | |
| 1785 | lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF |
| 1786 | |
| 1787 | The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags |
| 1788 | pair provide additional information particular to the objects they represent. |
| 1789 | |
| 1790 | Eventfd files |
| 1791 | ~~~~~~~~~~~~~ |
| 1792 | pos: 0 |
| 1793 | flags: 04002 |
| 1794 | mnt_id: 9 |
| 1795 | eventfd-count: 5a |
| 1796 | |
| 1797 | where 'eventfd-count' is hex value of a counter. |
| 1798 | |
| 1799 | Signalfd files |
| 1800 | ~~~~~~~~~~~~~~ |
| 1801 | pos: 0 |
| 1802 | flags: 04002 |
| 1803 | mnt_id: 9 |
| 1804 | sigmask: 0000000000000200 |
| 1805 | |
| 1806 | where 'sigmask' is hex value of the signal mask associated |
| 1807 | with a file. |
| 1808 | |
| 1809 | Epoll files |
| 1810 | ~~~~~~~~~~~ |
| 1811 | pos: 0 |
| 1812 | flags: 02 |
| 1813 | mnt_id: 9 |
| 1814 | tfd: 5 events: 1d data: ffffffffffffffff pos:0 ino:61af sdev:7 |
| 1815 | |
| 1816 | where 'tfd' is a target file descriptor number in decimal form, |
| 1817 | 'events' is events mask being watched and the 'data' is data |
| 1818 | associated with a target [see epoll(7) for more details]. |
| 1819 | |
| 1820 | The 'pos' is current offset of the target file in decimal form |
| 1821 | [see lseek(2)], 'ino' and 'sdev' are inode and device numbers |
| 1822 | where target file resides, all in hex format. |
| 1823 | |
| 1824 | Fsnotify files |
| 1825 | ~~~~~~~~~~~~~~ |
| 1826 | For inotify files the format is the following |
| 1827 | |
| 1828 | pos: 0 |
| 1829 | flags: 02000000 |
| 1830 | inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d |
| 1831 | |
| 1832 | where 'wd' is a watch descriptor in decimal form, ie a target file |
| 1833 | descriptor number, 'ino' and 'sdev' are inode and device where the |
| 1834 | target file resides and the 'mask' is the mask of events, all in hex |
| 1835 | form [see inotify(7) for more details]. |
| 1836 | |
| 1837 | If the kernel was built with exportfs support, the path to the target |
| 1838 | file is encoded as a file handle. The file handle is provided by three |
| 1839 | fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex |
| 1840 | format. |
| 1841 | |
| 1842 | If the kernel is built without exportfs support the file handle won't be |
| 1843 | printed out. |
| 1844 | |
| 1845 | If there is no inotify mark attached yet the 'inotify' line will be omitted. |
| 1846 | |
| 1847 | For fanotify files the format is |
| 1848 | |
| 1849 | pos: 0 |
| 1850 | flags: 02 |
| 1851 | mnt_id: 9 |
| 1852 | fanotify flags:10 event-flags:0 |
| 1853 | fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003 |
| 1854 | fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4 |
| 1855 | |
| 1856 | where fanotify 'flags' and 'event-flags' are values used in fanotify_init |
| 1857 | call, 'mnt_id' is the mount point identifier, 'mflags' is the value of |
| 1858 | flags associated with mark which are tracked separately from events |
| 1859 | mask. 'ino', 'sdev' are target inode and device, 'mask' is the events |
| 1860 | mask and 'ignored_mask' is the mask of events which are to be ignored. |
| 1861 | All in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask' |
| 1862 | does provide information about flags and mask used in fanotify_mark |
| 1863 | call [see fsnotify manpage for details]. |
| 1864 | |
| 1865 | While the first three lines are mandatory and always printed, the rest is |
| 1866 | optional and may be omitted if no marks created yet. |
| 1867 | |
| 1868 | Timerfd files |
| 1869 | ~~~~~~~~~~~~~ |
| 1870 | |
| 1871 | pos: 0 |
| 1872 | flags: 02 |
| 1873 | mnt_id: 9 |
| 1874 | clockid: 0 |
| 1875 | ticks: 0 |
| 1876 | settime flags: 01 |
| 1877 | it_value: (0, 49406829) |
| 1878 | it_interval: (1, 0) |
| 1879 | |
| 1880 | where 'clockid' is the clock type and 'ticks' is the number of the timer expirations |
| 1881 | that have occurred [see timerfd_create(2) for details]. 'settime flags' are |
| 1882 | flags in octal form been used to setup the timer [see timerfd_settime(2) for |
| 1883 | details]. 'it_value' is remaining time until the timer exiration. |
| 1884 | 'it_interval' is the interval for the timer. Note the timer might be set up |
| 1885 | with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value' |
| 1886 | still exhibits timer's remaining time. |
| 1887 | |
| 1888 | 3.9 /proc/<pid>/map_files - Information about memory mapped files |
| 1889 | --------------------------------------------------------------------- |
| 1890 | This directory contains symbolic links which represent memory mapped files |
| 1891 | the process is maintaining. Example output: |
| 1892 | |
| 1893 | | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so |
| 1894 | | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so |
| 1895 | | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so |
| 1896 | | ... |
| 1897 | | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1 |
| 1898 | | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls |
| 1899 | |
| 1900 | The name of a link represents the virtual memory bounds of a mapping, i.e. |
| 1901 | vm_area_struct::vm_start-vm_area_struct::vm_end. |
| 1902 | |
| 1903 | The main purpose of the map_files is to retrieve a set of memory mapped |
| 1904 | files in a fast way instead of parsing /proc/<pid>/maps or |
| 1905 | /proc/<pid>/smaps, both of which contain many more records. At the same |
| 1906 | time one can open(2) mappings from the listings of two processes and |
| 1907 | comparing their inode numbers to figure out which anonymous memory areas |
| 1908 | are actually shared. |
| 1909 | |
| 1910 | 3.10 /proc/<pid>/timerslack_ns - Task timerslack value |
| 1911 | --------------------------------------------------------- |
| 1912 | This file provides the value of the task's timerslack value in nanoseconds. |
| 1913 | This value specifies a amount of time that normal timers may be deferred |
| 1914 | in order to coalesce timers and avoid unnecessary wakeups. |
| 1915 | |
| 1916 | This allows a task's interactivity vs power consumption trade off to be |
| 1917 | adjusted. |
| 1918 | |
| 1919 | Writing 0 to the file will set the tasks timerslack to the default value. |
| 1920 | |
| 1921 | Valid values are from 0 - ULLONG_MAX |
| 1922 | |
| 1923 | An application setting the value must have PTRACE_MODE_ATTACH_FSCREDS level |
| 1924 | permissions on the task specified to change its timerslack_ns value. |
| 1925 | |
| 1926 | 3.11 /proc/<pid>/patch_state - Livepatch patch operation state |
| 1927 | ----------------------------------------------------------------- |
| 1928 | When CONFIG_LIVEPATCH is enabled, this file displays the value of the |
| 1929 | patch state for the task. |
| 1930 | |
| 1931 | A value of '-1' indicates that no patch is in transition. |
| 1932 | |
| 1933 | A value of '0' indicates that a patch is in transition and the task is |
| 1934 | unpatched. If the patch is being enabled, then the task hasn't been |
| 1935 | patched yet. If the patch is being disabled, then the task has already |
| 1936 | been unpatched. |
| 1937 | |
| 1938 | A value of '1' indicates that a patch is in transition and the task is |
| 1939 | patched. If the patch is being enabled, then the task has already been |
| 1940 | patched. If the patch is being disabled, then the task hasn't been |
| 1941 | unpatched yet. |
| 1942 | |
| 1943 | |
| 1944 | ------------------------------------------------------------------------------ |
| 1945 | Configuring procfs |
| 1946 | ------------------------------------------------------------------------------ |
| 1947 | |
| 1948 | 4.1 Mount options |
| 1949 | --------------------- |
| 1950 | |
| 1951 | The following mount options are supported: |
| 1952 | |
| 1953 | hidepid= Set /proc/<pid>/ access mode. |
| 1954 | gid= Set the group authorized to learn processes information. |
| 1955 | |
| 1956 | hidepid=0 means classic mode - everybody may access all /proc/<pid>/ directories |
| 1957 | (default). |
| 1958 | |
| 1959 | hidepid=1 means users may not access any /proc/<pid>/ directories but their |
| 1960 | own. Sensitive files like cmdline, sched*, status are now protected against |
| 1961 | other users. This makes it impossible to learn whether any user runs |
| 1962 | specific program (given the program doesn't reveal itself by its behaviour). |
| 1963 | As an additional bonus, as /proc/<pid>/cmdline is unaccessible for other users, |
| 1964 | poorly written programs passing sensitive information via program arguments are |
| 1965 | now protected against local eavesdroppers. |
| 1966 | |
| 1967 | hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be fully invisible to other |
| 1968 | users. It doesn't mean that it hides a fact whether a process with a specific |
| 1969 | pid value exists (it can be learned by other means, e.g. by "kill -0 $PID"), |
| 1970 | but it hides process' uid and gid, which may be learned by stat()'ing |
| 1971 | /proc/<pid>/ otherwise. It greatly complicates an intruder's task of gathering |
| 1972 | information about running processes, whether some daemon runs with elevated |
| 1973 | privileges, whether other user runs some sensitive program, whether other users |
| 1974 | run any program at all, etc. |
| 1975 | |
| 1976 | gid= defines a group authorized to learn processes information otherwise |
| 1977 | prohibited by hidepid=. If you use some daemon like identd which needs to learn |
| 1978 | information about processes information, just add identd to this group. |