1------------------------------------------------------------------------------ 2 T H E /proc F I L E S Y S T E M 3------------------------------------------------------------------------------ 4/proc/sys Terrehon Bowden <terrehon@pacbell.net> October 7 1999 5 Bodo Bauer <bb@ricochet.net> 6 72.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000 8move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009 9------------------------------------------------------------------------------ 10Version 1.3 Kernel version 2.2.12 11 Kernel version 2.4.0-test11-pre4 12------------------------------------------------------------------------------ 13fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009 14 15Table of Contents 16----------------- 17 18 0 Preface 19 0.1 Introduction/Credits 20 0.2 Legal Stuff 21 22 1 Collecting System Information 23 1.1 Process-Specific Subdirectories 24 1.2 Kernel data 25 1.3 IDE devices in /proc/ide 26 1.4 Networking info in /proc/net 27 1.5 SCSI info 28 1.6 Parallel port info in /proc/parport 29 1.7 TTY info in /proc/tty 30 1.8 Miscellaneous kernel statistics in /proc/stat 31 1.9 Ext4 file system parameters 32 33 2 Modifying System Parameters 34 35 3 Per-Process Parameters 36 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj - Adjust the oom-killer 37 score 38 3.2 /proc/<pid>/oom_score - Display current oom-killer score 39 3.3 /proc/<pid>/io - Display the IO accounting fields 40 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings 41 3.5 /proc/<pid>/mountinfo - Information about mounts 42 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 43 44 45------------------------------------------------------------------------------ 46Preface 47------------------------------------------------------------------------------ 48 490.1 Introduction/Credits 50------------------------ 51 52This documentation is part of a soon (or so we hope) to be released book on 53the SuSE Linux distribution. As there is no complete documentation for the 54/proc file system and we've used many freely available sources to write these 55chapters, it seems only fair to give the work back to the Linux community. 56This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm 57afraid it's still far from complete, but we hope it will be useful. As far as 58we know, it is the first 'all-in-one' document about the /proc file system. It 59is focused on the Intel x86 hardware, so if you are looking for PPC, ARM, 60SPARC, AXP, etc., features, you probably won't find what you are looking for. 61It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But 62additions and patches are welcome and will be added to this document if you 63mail them to Bodo. 64 65We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of 66other people for help compiling this documentation. We'd also like to extend a 67special thank you to Andi Kleen for documentation, which we relied on heavily 68to create this document, as well as the additional information he provided. 69Thanks to everybody else who contributed source or docs to the Linux kernel 70and helped create a great piece of software... :) 71 72If you have any comments, corrections or additions, please don't hesitate to 73contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this 74document. 75 76The latest version of this document is available online at 77http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html 78 79If the above direction does not works for you, you could try the kernel 80mailing list at linux-kernel@vger.kernel.org and/or try to reach me at 81comandante@zaralinux.com. 82 830.2 Legal Stuff 84--------------- 85 86We don't guarantee the correctness of this document, and if you come to us 87complaining about how you screwed up your system because of incorrect 88documentation, we won't feel responsible... 89 90------------------------------------------------------------------------------ 91CHAPTER 1: COLLECTING SYSTEM INFORMATION 92------------------------------------------------------------------------------ 93 94------------------------------------------------------------------------------ 95In This Chapter 96------------------------------------------------------------------------------ 97* Investigating the properties of the pseudo file system /proc and its 98 ability to provide information on the running Linux system 99* Examining /proc's structure 100* Uncovering various information about the kernel and the processes running 101 on the system 102------------------------------------------------------------------------------ 103 104 105The proc file system acts as an interface to internal data structures in the 106kernel. It can be used to obtain information about the system and to change 107certain kernel parameters at runtime (sysctl). 108 109First, we'll take a look at the read-only parts of /proc. In Chapter 2, we 110show you how you can use /proc/sys to change settings. 111 1121.1 Process-Specific Subdirectories 113----------------------------------- 114 115The directory /proc contains (among other things) one subdirectory for each 116process running on the system, which is named after the process ID (PID). 117 118The link self points to the process reading the file system. Each process 119subdirectory has the entries listed in Table 1-1. 120 121 122Table 1-1: Process specific entries in /proc 123.............................................................................. 124 File Content 125 clear_refs Clears page referenced bits shown in smaps output 126 cmdline Command line arguments 127 cpu Current and last cpu in which it was executed (2.4)(smp) 128 cwd Link to the current working directory 129 environ Values of environment variables 130 exe Link to the executable of this process 131 fd Directory, which contains all file descriptors 132 maps Memory maps to executables and library files (2.4) 133 mem Memory held by this process 134 root Link to the root directory of this process 135 stat Process status 136 statm Process memory status information 137 status Process status in human readable form 138 wchan If CONFIG_KALLSYMS is set, a pre-decoded wchan 139 stack Report full stack trace, enable via CONFIG_STACKTRACE 140 smaps a extension based on maps, showing the memory consumption of 141 each mapping 142.............................................................................. 143 144For example, to get the status information of a process, all you have to do is 145read the file /proc/PID/status: 146 147 >cat /proc/self/status 148 Name: cat 149 State: R (running) 150 Tgid: 5452 151 Pid: 5452 152 PPid: 743 153 TracerPid: 0 (2.4) 154 Uid: 501 501 501 501 155 Gid: 100 100 100 100 156 FDSize: 256 157 Groups: 100 14 16 158 VmPeak: 5004 kB 159 VmSize: 5004 kB 160 VmLck: 0 kB 161 VmHWM: 476 kB 162 VmRSS: 476 kB 163 VmData: 156 kB 164 VmStk: 88 kB 165 VmExe: 68 kB 166 VmLib: 1412 kB 167 VmPTE: 20 kb 168 VmSwap: 0 kB 169 Threads: 1 170 SigQ: 0/28578 171 SigPnd: 0000000000000000 172 ShdPnd: 0000000000000000 173 SigBlk: 0000000000000000 174 SigIgn: 0000000000000000 175 SigCgt: 0000000000000000 176 CapInh: 00000000fffffeff 177 CapPrm: 0000000000000000 178 CapEff: 0000000000000000 179 CapBnd: ffffffffffffffff 180 voluntary_ctxt_switches: 0 181 nonvoluntary_ctxt_switches: 1 182 183This shows you nearly the same information you would get if you viewed it with 184the ps command. In fact, ps uses the proc file system to obtain its 185information. But you get a more detailed view of the process by reading the 186file /proc/PID/status. It fields are described in table 1-2. 187 188The statm file contains more detailed information about the process 189memory usage. Its seven fields are explained in Table 1-3. The stat file 190contains details information about the process itself. Its fields are 191explained in Table 1-4. 192 193(for SMP CONFIG users) 194For making accounting scalable, RSS related information are handled in 195asynchronous manner and the vaule may not be very precise. To see a precise 196snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table. 197It's slow but very precise. 198 199Table 1-2: Contents of the status files (as of 2.6.30-rc7) 200.............................................................................. 201 Field Content 202 Name filename of the executable 203 State state (R is running, S is sleeping, D is sleeping 204 in an uninterruptible wait, Z is zombie, 205 T is traced or stopped) 206 Tgid thread group ID 207 Pid process id 208 PPid process id of the parent process 209 TracerPid PID of process tracing this process (0 if not) 210 Uid Real, effective, saved set, and file system UIDs 211 Gid Real, effective, saved set, and file system GIDs 212 FDSize number of file descriptor slots currently allocated 213 Groups supplementary group list 214 VmPeak peak virtual memory size 215 VmSize total program size 216 VmLck locked memory size 217 VmHWM peak resident set size ("high water mark") 218 VmRSS size of memory portions 219 VmData size of data, stack, and text segments 220 VmStk size of data, stack, and text segments 221 VmExe size of text segment 222 VmLib size of shared library code 223 VmPTE size of page table entries 224 VmSwap size of swap usage (the number of referred swapents) 225 Threads number of threads 226 SigQ number of signals queued/max. number for queue 227 SigPnd bitmap of pending signals for the thread 228 ShdPnd bitmap of shared pending signals for the process 229 SigBlk bitmap of blocked signals 230 SigIgn bitmap of ignored signals 231 SigCgt bitmap of catched signals 232 CapInh bitmap of inheritable capabilities 233 CapPrm bitmap of permitted capabilities 234 CapEff bitmap of effective capabilities 235 CapBnd bitmap of capabilities bounding set 236 Cpus_allowed mask of CPUs on which this process may run 237 Cpus_allowed_list Same as previous, but in "list format" 238 Mems_allowed mask of memory nodes allowed to this process 239 Mems_allowed_list Same as previous, but in "list format" 240 voluntary_ctxt_switches number of voluntary context switches 241 nonvoluntary_ctxt_switches number of non voluntary context switches 242.............................................................................. 243 244Table 1-3: Contents of the statm files (as of 2.6.8-rc3) 245.............................................................................. 246 Field Content 247 size total program size (pages) (same as VmSize in status) 248 resident size of memory portions (pages) (same as VmRSS in status) 249 shared number of pages that are shared (i.e. backed by a file) 250 trs number of pages that are 'code' (not including libs; broken, 251 includes data segment) 252 lrs number of pages of library (always 0 on 2.6) 253 drs number of pages of data/stack (including libs; broken, 254 includes library text) 255 dt number of dirty pages (always 0 on 2.6) 256.............................................................................. 257 258 259Table 1-4: Contents of the stat files (as of 2.6.30-rc7) 260.............................................................................. 261 Field Content 262 pid process id 263 tcomm filename of the executable 264 state state (R is running, S is sleeping, D is sleeping in an 265 uninterruptible wait, Z is zombie, T is traced or stopped) 266 ppid process id of the parent process 267 pgrp pgrp of the process 268 sid session id 269 tty_nr tty the process uses 270 tty_pgrp pgrp of the tty 271 flags task flags 272 min_flt number of minor faults 273 cmin_flt number of minor faults with child's 274 maj_flt number of major faults 275 cmaj_flt number of major faults with child's 276 utime user mode jiffies 277 stime kernel mode jiffies 278 cutime user mode jiffies with child's 279 cstime kernel mode jiffies with child's 280 priority priority level 281 nice nice level 282 num_threads number of threads 283 it_real_value (obsolete, always 0) 284 start_time time the process started after system boot 285 vsize virtual memory size 286 rss resident set memory size 287 rsslim current limit in bytes on the rss 288 start_code address above which program text can run 289 end_code address below which program text can run 290 start_stack address of the start of the stack 291 esp current value of ESP 292 eip current value of EIP 293 pending bitmap of pending signals 294 blocked bitmap of blocked signals 295 sigign bitmap of ignored signals 296 sigcatch bitmap of catched signals 297 wchan address where process went to sleep 298 0 (place holder) 299 0 (place holder) 300 exit_signal signal to send to parent thread on exit 301 task_cpu which CPU the task is scheduled on 302 rt_priority realtime priority 303 policy scheduling policy (man sched_setscheduler) 304 blkio_ticks time spent waiting for block IO 305 gtime guest time of the task in jiffies 306 cgtime guest time of the task children in jiffies 307.............................................................................. 308 309The /proc/PID/maps file containing the currently mapped memory regions and 310their access permissions. 311 312The format is: 313 314address perms offset dev inode pathname 315 31608048000-08049000 r-xp 00000000 03:00 8312 /opt/test 31708049000-0804a000 rw-p 00001000 03:00 8312 /opt/test 3180804a000-0806b000 rw-p 00000000 00:00 0 [heap] 319a7cb1000-a7cb2000 ---p 00000000 00:00 0 320a7cb2000-a7eb2000 rw-p 00000000 00:00 0 321a7eb2000-a7eb3000 ---p 00000000 00:00 0 322a7eb3000-a7ed5000 rw-p 00000000 00:00 0 323a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6 324a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6 325a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6 326a800b000-a800e000 rw-p 00000000 00:00 0 327a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0 328a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0 329a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0 330a8024000-a8027000 rw-p 00000000 00:00 0 331a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2 332a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2 333a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2 334aff35000-aff4a000 rw-p 00000000 00:00 0 [stack] 335ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso] 336 337where "address" is the address space in the process that it occupies, "perms" 338is a set of permissions: 339 340 r = read 341 w = write 342 x = execute 343 s = shared 344 p = private (copy on write) 345 346"offset" is the offset into the mapping, "dev" is the device (major:minor), and 347"inode" is the inode on that device. 0 indicates that no inode is associated 348with the memory region, as the case would be with BSS (uninitialized data). 349The "pathname" shows the name associated file for this mapping. If the mapping 350is not associated with a file: 351 352 [heap] = the heap of the program 353 [stack] = the stack of the main process 354 [vdso] = the "virtual dynamic shared object", 355 the kernel system call handler 356 357 or if empty, the mapping is anonymous. 358 359 360The /proc/PID/smaps is an extension based on maps, showing the memory 361consumption for each of the process's mappings. For each of mappings there 362is a series of lines such as the following: 363 36408048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash 365Size: 1084 kB 366Rss: 892 kB 367Pss: 374 kB 368Shared_Clean: 892 kB 369Shared_Dirty: 0 kB 370Private_Clean: 0 kB 371Private_Dirty: 0 kB 372Referenced: 892 kB 373Swap: 0 kB 374KernelPageSize: 4 kB 375MMUPageSize: 4 kB 376 377The first of these lines shows the same information as is displayed for the 378mapping in /proc/PID/maps. The remaining lines show the size of the mapping, 379the amount of the mapping that is currently resident in RAM, the "proportional 380set size��� (divide each shared page by the number of processes sharing it), the 381number of clean and dirty shared pages in the mapping, and the number of clean 382and dirty private pages in the mapping. The "Referenced" indicates the amount 383of memory currently marked as referenced or accessed. 384 385This file is only present if the CONFIG_MMU kernel configuration option is 386enabled. 387 388The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG 389bits on both physical and virtual pages associated with a process. 390To clear the bits for all the pages associated with the process 391 > echo 1 > /proc/PID/clear_refs 392 393To clear the bits for the anonymous pages associated with the process 394 > echo 2 > /proc/PID/clear_refs 395 396To clear the bits for the file mapped pages associated with the process 397 > echo 3 > /proc/PID/clear_refs 398Any other value written to /proc/PID/clear_refs will have no effect. 399 400 4011.2 Kernel data 402--------------- 403 404Similar to the process entries, the kernel data files give information about 405the running kernel. The files used to obtain this information are contained in 406/proc and are listed in Table 1-5. Not all of these will be present in your 407system. It depends on the kernel configuration and the loaded modules, which 408files are there, and which are missing. 409 410Table 1-5: Kernel info in /proc 411.............................................................................. 412 File Content 413 apm Advanced power management info 414 buddyinfo Kernel memory allocator information (see text) (2.5) 415 bus Directory containing bus specific information 416 cmdline Kernel command line 417 cpuinfo Info about the CPU 418 devices Available devices (block and character) 419 dma Used DMS channels 420 filesystems Supported filesystems 421 driver Various drivers grouped here, currently rtc (2.4) 422 execdomains Execdomains, related to security (2.4) 423 fb Frame Buffer devices (2.4) 424 fs File system parameters, currently nfs/exports (2.4) 425 ide Directory containing info about the IDE subsystem 426 interrupts Interrupt usage 427 iomem Memory map (2.4) 428 ioports I/O port usage 429 irq Masks for irq to cpu affinity (2.4)(smp?) 430 isapnp ISA PnP (Plug&Play) Info (2.4) 431 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4)) 432 kmsg Kernel messages 433 ksyms Kernel symbol table 434 loadavg Load average of last 1, 5 & 15 minutes 435 locks Kernel locks 436 meminfo Memory info 437 misc Miscellaneous 438 modules List of loaded modules 439 mounts Mounted filesystems 440 net Networking info (see text) 441 pagetypeinfo Additional page allocator information (see text) (2.5) 442 partitions Table of partitions known to the system 443 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/, 444 decoupled by lspci (2.4) 445 rtc Real time clock 446 scsi SCSI info (see text) 447 slabinfo Slab pool info 448 softirqs softirq usage 449 stat Overall statistics 450 swaps Swap space utilization 451 sys See chapter 2 452 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4) 453 tty Info of tty drivers 454 uptime System uptime 455 version Kernel version 456 video bttv info of video resources (2.4) 457 vmallocinfo Show vmalloced areas 458.............................................................................. 459 460You can, for example, check which interrupts are currently in use and what 461they are used for by looking in the file /proc/interrupts: 462 463 > cat /proc/interrupts 464 CPU0 465 0: 8728810 XT-PIC timer 466 1: 895 XT-PIC keyboard 467 2: 0 XT-PIC cascade 468 3: 531695 XT-PIC aha152x 469 4: 2014133 XT-PIC serial 470 5: 44401 XT-PIC pcnet_cs 471 8: 2 XT-PIC rtc 472 11: 8 XT-PIC i82365 473 12: 182918 XT-PIC PS/2 Mouse 474 13: 1 XT-PIC fpu 475 14: 1232265 XT-PIC ide0 476 15: 7 XT-PIC ide1 477 NMI: 0 478 479In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the 480output of a SMP machine): 481 482 > cat /proc/interrupts 483 484 CPU0 CPU1 485 0: 1243498 1214548 IO-APIC-edge timer 486 1: 8949 8958 IO-APIC-edge keyboard 487 2: 0 0 XT-PIC cascade 488 5: 11286 10161 IO-APIC-edge soundblaster 489 8: 1 0 IO-APIC-edge rtc 490 9: 27422 27407 IO-APIC-edge 3c503 491 12: 113645 113873 IO-APIC-edge PS/2 Mouse 492 13: 0 0 XT-PIC fpu 493 14: 22491 24012 IO-APIC-edge ide0 494 15: 2183 2415 IO-APIC-edge ide1 495 17: 30564 30414 IO-APIC-level eth0 496 18: 177 164 IO-APIC-level bttv 497 NMI: 2457961 2457959 498 LOC: 2457882 2457881 499 ERR: 2155 500 501NMI is incremented in this case because every timer interrupt generates a NMI 502(Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups. 503 504LOC is the local interrupt counter of the internal APIC of every CPU. 505 506ERR is incremented in the case of errors in the IO-APIC bus (the bus that 507connects the CPUs in a SMP system. This means that an error has been detected, 508the IO-APIC automatically retry the transmission, so it should not be a big 509problem, but you should read the SMP-FAQ. 510 511In 2.6.2* /proc/interrupts was expanded again. This time the goal was for 512/proc/interrupts to display every IRQ vector in use by the system, not 513just those considered 'most important'. The new vectors are: 514 515 THR -- interrupt raised when a machine check threshold counter 516 (typically counting ECC corrected errors of memory or cache) exceeds 517 a configurable threshold. Only available on some systems. 518 519 TRM -- a thermal event interrupt occurs when a temperature threshold 520 has been exceeded for the CPU. This interrupt may also be generated 521 when the temperature drops back to normal. 522 523 SPU -- a spurious interrupt is some interrupt that was raised then lowered 524 by some IO device before it could be fully processed by the APIC. Hence 525 the APIC sees the interrupt but does not know what device it came from. 526 For this case the APIC will generate the interrupt with a IRQ vector 527 of 0xff. This might also be generated by chipset bugs. 528 529 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are 530 sent from one CPU to another per the needs of the OS. Typically, 531 their statistics are used by kernel developers and interested users to 532 determine the occurrence of interrupts of the given type. 533 534The above IRQ vectors are displayed only when relevent. For example, 535the threshold vector does not exist on x86_64 platforms. Others are 536suppressed when the system is a uniprocessor. As of this writing, only 537i386 and x86_64 platforms support the new IRQ vector displays. 538 539Of some interest is the introduction of the /proc/irq directory to 2.4. 540It could be used to set IRQ to CPU affinity, this means that you can "hook" an 541IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the 542irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and 543prof_cpu_mask. 544 545For example 546 > ls /proc/irq/ 547 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask 548 1 11 13 15 17 19 3 5 7 9 default_smp_affinity 549 > ls /proc/irq/0/ 550 smp_affinity 551 552smp_affinity is a bitmask, in which you can specify which CPUs can handle the 553IRQ, you can set it by doing: 554 555 > echo 1 > /proc/irq/10/smp_affinity 556 557This means that only the first CPU will handle the IRQ, but you can also echo 5585 which means that only the first and fourth CPU can handle the IRQ. 559 560The contents of each smp_affinity file is the same by default: 561 562 > cat /proc/irq/0/smp_affinity 563 ffffffff 564 565The default_smp_affinity mask applies to all non-active IRQs, which are the 566IRQs which have not yet been allocated/activated, and hence which lack a 567/proc/irq/[0-9]* directory. 568 569The node file on an SMP system shows the node to which the device using the IRQ 570reports itself as being attached. This hardware locality information does not 571include information about any possible driver locality preference. 572 573prof_cpu_mask specifies which CPUs are to be profiled by the system wide 574profiler. Default value is ffffffff (all cpus). 575 576The way IRQs are routed is handled by the IO-APIC, and it's Round Robin 577between all the CPUs which are allowed to handle it. As usual the kernel has 578more info than you and does a better job than you, so the defaults are the 579best choice for almost everyone. 580 581There are three more important subdirectories in /proc: net, scsi, and sys. 582The general rule is that the contents, or even the existence of these 583directories, depend on your kernel configuration. If SCSI is not enabled, the 584directory scsi may not exist. The same is true with the net, which is there 585only when networking support is present in the running kernel. 586 587The slabinfo file gives information about memory usage at the slab level. 588Linux uses slab pools for memory management above page level in version 2.2. 589Commonly used objects have their own slab pool (such as network buffers, 590directory cache, and so on). 591 592.............................................................................. 593 594> cat /proc/buddyinfo 595 596Node 0, zone DMA 0 4 5 4 4 3 ... 597Node 0, zone Normal 1 0 0 1 101 8 ... 598Node 0, zone HighMem 2 0 0 1 1 0 ... 599 600External fragmentation is a problem under some workloads, and buddyinfo is a 601useful tool for helping diagnose these problems. Buddyinfo will give you a 602clue as to how big an area you can safely allocate, or why a previous 603allocation failed. 604 605Each column represents the number of pages of a certain order which are 606available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in 607ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE 608available in ZONE_NORMAL, etc... 609 610More information relevant to external fragmentation can be found in 611pagetypeinfo. 612 613> cat /proc/pagetypeinfo 614Page block order: 9 615Pages per block: 512 616 617Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10 618Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0 619Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0 620Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2 621Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0 622Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0 623Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9 624Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0 625Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452 626Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0 627Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0 628 629Number of blocks type Unmovable Reclaimable Movable Reserve Isolate 630Node 0, zone DMA 2 0 5 1 0 631Node 0, zone DMA32 41 6 967 2 0 632 633Fragmentation avoidance in the kernel works by grouping pages of different 634migrate types into the same contiguous regions of memory called page blocks. 635A page block is typically the size of the default hugepage size e.g. 2MB on 636X86-64. By keeping pages grouped based on their ability to move, the kernel 637can reclaim pages within a page block to satisfy a high-order allocation. 638 639The pagetypinfo begins with information on the size of a page block. It 640then gives the same type of information as buddyinfo except broken down 641by migrate-type and finishes with details on how many page blocks of each 642type exist. 643 644If min_free_kbytes has been tuned correctly (recommendations made by hugeadm 645from libhugetlbfs http://sourceforge.net/projects/libhugetlbfs/), one can 646make an estimate of the likely number of huge pages that can be allocated 647at a given point in time. All the "Movable" blocks should be allocatable 648unless memory has been mlock()'d. Some of the Reclaimable blocks should 649also be allocatable although a lot of filesystem metadata may have to be 650reclaimed to achieve this. 651 652.............................................................................. 653 654meminfo: 655 656Provides information about distribution and utilization of memory. This 657varies by architecture and compile options. The following is from a 65816GB PIII, which has highmem enabled. You may not have all of these fields. 659 660> cat /proc/meminfo 661 662 663MemTotal: 16344972 kB 664MemFree: 13634064 kB 665Buffers: 3656 kB 666Cached: 1195708 kB 667SwapCached: 0 kB 668Active: 891636 kB 669Inactive: 1077224 kB 670HighTotal: 15597528 kB 671HighFree: 13629632 kB 672LowTotal: 747444 kB 673LowFree: 4432 kB 674SwapTotal: 0 kB 675SwapFree: 0 kB 676Dirty: 968 kB 677Writeback: 0 kB 678AnonPages: 861800 kB 679Mapped: 280372 kB 680Slab: 284364 kB 681SReclaimable: 159856 kB 682SUnreclaim: 124508 kB 683PageTables: 24448 kB 684NFS_Unstable: 0 kB 685Bounce: 0 kB 686WritebackTmp: 0 kB 687CommitLimit: 7669796 kB 688Committed_AS: 100056 kB 689VmallocTotal: 112216 kB 690VmallocUsed: 428 kB 691VmallocChunk: 111088 kB 692 693 MemTotal: Total usable ram (i.e. physical ram minus a few reserved 694 bits and the kernel binary code) 695 MemFree: The sum of LowFree+HighFree 696 Buffers: Relatively temporary storage for raw disk blocks 697 shouldn't get tremendously large (20MB or so) 698 Cached: in-memory cache for files read from the disk (the 699 pagecache). Doesn't include SwapCached 700 SwapCached: Memory that once was swapped out, is swapped back in but 701 still also is in the swapfile (if memory is needed it 702 doesn't need to be swapped out AGAIN because it is already 703 in the swapfile. This saves I/O) 704 Active: Memory that has been used more recently and usually not 705 reclaimed unless absolutely necessary. 706 Inactive: Memory which has been less recently used. It is more 707 eligible to be reclaimed for other purposes 708 HighTotal: 709 HighFree: Highmem is all memory above ~860MB of physical memory 710 Highmem areas are for use by userspace programs, or 711 for the pagecache. The kernel must use tricks to access 712 this memory, making it slower to access than lowmem. 713 LowTotal: 714 LowFree: Lowmem is memory which can be used for everything that 715 highmem can be used for, but it is also available for the 716 kernel's use for its own data structures. Among many 717 other things, it is where everything from the Slab is 718 allocated. Bad things happen when you're out of lowmem. 719 SwapTotal: total amount of swap space available 720 SwapFree: Memory which has been evicted from RAM, and is temporarily 721 on the disk 722 Dirty: Memory which is waiting to get written back to the disk 723 Writeback: Memory which is actively being written back to the disk 724 AnonPages: Non-file backed pages mapped into userspace page tables 725 Mapped: files which have been mmaped, such as libraries 726 Slab: in-kernel data structures cache 727SReclaimable: Part of Slab, that might be reclaimed, such as caches 728 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure 729 PageTables: amount of memory dedicated to the lowest level of page 730 tables. 731NFS_Unstable: NFS pages sent to the server, but not yet committed to stable 732 storage 733 Bounce: Memory used for block device "bounce buffers" 734WritebackTmp: Memory used by FUSE for temporary writeback buffers 735 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'), 736 this is the total amount of memory currently available to 737 be allocated on the system. This limit is only adhered to 738 if strict overcommit accounting is enabled (mode 2 in 739 'vm.overcommit_memory'). 740 The CommitLimit is calculated with the following formula: 741 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap 742 For example, on a system with 1G of physical RAM and 7G 743 of swap with a `vm.overcommit_ratio` of 30 it would 744 yield a CommitLimit of 7.3G. 745 For more details, see the memory overcommit documentation 746 in vm/overcommit-accounting. 747Committed_AS: The amount of memory presently allocated on the system. 748 The committed memory is a sum of all of the memory which 749 has been allocated by processes, even if it has not been 750 "used" by them as of yet. A process which malloc()'s 1G 751 of memory, but only touches 300M of it will only show up 752 as using 300M of memory even if it has the address space 753 allocated for the entire 1G. This 1G is memory which has 754 been "committed" to by the VM and can be used at any time 755 by the allocating application. With strict overcommit 756 enabled on the system (mode 2 in 'vm.overcommit_memory'), 757 allocations which would exceed the CommitLimit (detailed 758 above) will not be permitted. This is useful if one needs 759 to guarantee that processes will not fail due to lack of 760 memory once that memory has been successfully allocated. 761VmallocTotal: total size of vmalloc memory area 762 VmallocUsed: amount of vmalloc area which is used 763VmallocChunk: largest contiguous block of vmalloc area which is free 764 765.............................................................................. 766 767vmallocinfo: 768 769Provides information about vmalloced/vmaped areas. One line per area, 770containing the virtual address range of the area, size in bytes, 771caller information of the creator, and optional information depending 772on the kind of area : 773 774 pages=nr number of pages 775 phys=addr if a physical address was specified 776 ioremap I/O mapping (ioremap() and friends) 777 vmalloc vmalloc() area 778 vmap vmap()ed pages 779 user VM_USERMAP area 780 vpages buffer for pages pointers was vmalloced (huge area) 781 N<node>=nr (Only on NUMA kernels) 782 Number of pages allocated on memory node <node> 783 784> cat /proc/vmallocinfo 7850xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ... 786 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128 7870xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ... 788 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64 7890xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f... 790 phys=7fee8000 ioremap 7910xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f... 792 phys=7fee7000 ioremap 7930xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210 7940xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ... 795 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3 7960xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ... 797 pages=2 vmalloc N1=2 7980xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ... 799 /0x130 [x_tables] pages=4 vmalloc N0=4 8000xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ... 801 pages=14 vmalloc N2=14 8020xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ... 803 pages=4 vmalloc N1=4 8040xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ... 805 pages=2 vmalloc N1=2 8060xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ... 807 pages=10 vmalloc N0=10 808 809.............................................................................. 810 811softirqs: 812 813Provides counts of softirq handlers serviced since boot time, for each cpu. 814 815> cat /proc/softirqs 816 CPU0 CPU1 CPU2 CPU3 817 HI: 0 0 0 0 818 TIMER: 27166 27120 27097 27034 819 NET_TX: 0 0 0 17 820 NET_RX: 42 0 0 39 821 BLOCK: 0 0 107 1121 822 TASKLET: 0 0 0 290 823 SCHED: 27035 26983 26971 26746 824 HRTIMER: 0 0 0 0 825 RCU: 1678 1769 2178 2250 826 827 8281.3 IDE devices in /proc/ide 829---------------------------- 830 831The subdirectory /proc/ide contains information about all IDE devices of which 832the kernel is aware. There is one subdirectory for each IDE controller, the 833file drivers and a link for each IDE device, pointing to the device directory 834in the controller specific subtree. 835 836The file drivers contains general information about the drivers used for the 837IDE devices: 838 839 > cat /proc/ide/drivers 840 ide-cdrom version 4.53 841 ide-disk version 1.08 842 843More detailed information can be found in the controller specific 844subdirectories. These are named ide0, ide1 and so on. Each of these 845directories contains the files shown in table 1-6. 846 847 848Table 1-6: IDE controller info in /proc/ide/ide? 849.............................................................................. 850 File Content 851 channel IDE channel (0 or 1) 852 config Configuration (only for PCI/IDE bridge) 853 mate Mate name 854 model Type/Chipset of IDE controller 855.............................................................................. 856 857Each device connected to a controller has a separate subdirectory in the 858controllers directory. The files listed in table 1-7 are contained in these 859directories. 860 861 862Table 1-7: IDE device information 863.............................................................................. 864 File Content 865 cache The cache 866 capacity Capacity of the medium (in 512Byte blocks) 867 driver driver and version 868 geometry physical and logical geometry 869 identify device identify block 870 media media type 871 model device identifier 872 settings device setup 873 smart_thresholds IDE disk management thresholds 874 smart_values IDE disk management values 875.............................................................................. 876 877The most interesting file is settings. This file contains a nice overview of 878the drive parameters: 879 880 # cat /proc/ide/ide0/hda/settings 881 name value min max mode 882 ---- ----- --- --- ---- 883 bios_cyl 526 0 65535 rw 884 bios_head 255 0 255 rw 885 bios_sect 63 0 63 rw 886 breada_readahead 4 0 127 rw 887 bswap 0 0 1 r 888 file_readahead 72 0 2097151 rw 889 io_32bit 0 0 3 rw 890 keepsettings 0 0 1 rw 891 max_kb_per_request 122 1 127 rw 892 multcount 0 0 8 rw 893 nice1 1 0 1 rw 894 nowerr 0 0 1 rw 895 pio_mode write-only 0 255 w 896 slow 0 0 1 rw 897 unmaskirq 0 0 1 rw 898 using_dma 0 0 1 rw 899 900 9011.4 Networking info in /proc/net 902-------------------------------- 903 904The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the 905additional values you get for IP version 6 if you configure the kernel to 906support this. Table 1-9 lists the files and their meaning. 907 908 909Table 1-8: IPv6 info in /proc/net 910.............................................................................. 911 File Content 912 udp6 UDP sockets (IPv6) 913 tcp6 TCP sockets (IPv6) 914 raw6 Raw device statistics (IPv6) 915 igmp6 IP multicast addresses, which this host joined (IPv6) 916 if_inet6 List of IPv6 interface addresses 917 ipv6_route Kernel routing table for IPv6 918 rt6_stats Global IPv6 routing tables statistics 919 sockstat6 Socket statistics (IPv6) 920 snmp6 Snmp data (IPv6) 921.............................................................................. 922 923 924Table 1-9: Network info in /proc/net 925.............................................................................. 926 File Content 927 arp Kernel ARP table 928 dev network devices with statistics 929 dev_mcast the Layer2 multicast groups a device is listening too 930 (interface index, label, number of references, number of bound 931 addresses). 932 dev_stat network device status 933 ip_fwchains Firewall chain linkage 934 ip_fwnames Firewall chain names 935 ip_masq Directory containing the masquerading tables 936 ip_masquerade Major masquerading table 937 netstat Network statistics 938 raw raw device statistics 939 route Kernel routing table 940 rpc Directory containing rpc info 941 rt_cache Routing cache 942 snmp SNMP data 943 sockstat Socket statistics 944 tcp TCP sockets 945 tr_rif Token ring RIF routing table 946 udp UDP sockets 947 unix UNIX domain sockets 948 wireless Wireless interface data (Wavelan etc) 949 igmp IP multicast addresses, which this host joined 950 psched Global packet scheduler parameters. 951 netlink List of PF_NETLINK sockets 952 ip_mr_vifs List of multicast virtual interfaces 953 ip_mr_cache List of multicast routing cache 954.............................................................................. 955 956You can use this information to see which network devices are available in 957your system and how much traffic was routed over those devices: 958 959 > cat /proc/net/dev 960 Inter-|Receive |[... 961 face |bytes packets errs drop fifo frame compressed multicast|[... 962 lo: 908188 5596 0 0 0 0 0 0 [... 963 ppp0:15475140 20721 410 0 0 410 0 0 [... 964 eth0: 614530 7085 0 0 0 0 0 1 [... 965 966 ...] Transmit 967 ...] bytes packets errs drop fifo colls carrier compressed 968 ...] 908188 5596 0 0 0 0 0 0 969 ...] 1375103 17405 0 0 0 0 0 0 970 ...] 1703981 5535 0 0 0 3 0 0 971 972In addition, each Channel Bond interface has its own directory. For 973example, the bond0 device will have a directory called /proc/net/bond0/. 974It will contain information that is specific to that bond, such as the 975current slaves of the bond, the link status of the slaves, and how 976many times the slaves link has failed. 977 9781.5 SCSI info 979------------- 980 981If you have a SCSI host adapter in your system, you'll find a subdirectory 982named after the driver for this adapter in /proc/scsi. You'll also see a list 983of all recognized SCSI devices in /proc/scsi: 984 985 >cat /proc/scsi/scsi 986 Attached devices: 987 Host: scsi0 Channel: 00 Id: 00 Lun: 00 988 Vendor: IBM Model: DGHS09U Rev: 03E0 989 Type: Direct-Access ANSI SCSI revision: 03 990 Host: scsi0 Channel: 00 Id: 06 Lun: 00 991 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04 992 Type: CD-ROM ANSI SCSI revision: 02 993 994 995The directory named after the driver has one file for each adapter found in 996the system. These files contain information about the controller, including 997the used IRQ and the IO address range. The amount of information shown is 998dependent on the adapter you use. The example shows the output for an Adaptec 999AHA-2940 SCSI adapter: 1000 1001 > cat /proc/scsi/aic7xxx/0 1002 1003 Adaptec AIC7xxx driver version: 5.1.19/3.2.4 1004 Compile Options: 1005 TCQ Enabled By Default : Disabled 1006 AIC7XXX_PROC_STATS : Disabled 1007 AIC7XXX_RESET_DELAY : 5 1008 Adapter Configuration: 1009 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter 1010 Ultra Wide Controller 1011 PCI MMAPed I/O Base: 0xeb001000 1012 Adapter SEEPROM Config: SEEPROM found and used. 1013 Adaptec SCSI BIOS: Enabled 1014 IRQ: 10 1015 SCBs: Active 0, Max Active 2, 1016 Allocated 15, HW 16, Page 255 1017 Interrupts: 160328 1018 BIOS Control Word: 0x18b6 1019 Adapter Control Word: 0x005b 1020 Extended Translation: Enabled 1021 Disconnect Enable Flags: 0xffff 1022 Ultra Enable Flags: 0x0001 1023 Tag Queue Enable Flags: 0x0000 1024 Ordered Queue Tag Flags: 0x0000 1025 Default Tag Queue Depth: 8 1026 Tagged Queue By Device array for aic7xxx host instance 0: 1027 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255} 1028 Actual queue depth per device for aic7xxx host instance 0: 1029 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} 1030 Statistics: 1031 (scsi0:0:0:0) 1032 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8 1033 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0) 1034 Total transfers 160151 (74577 reads and 85574 writes) 1035 (scsi0:0:6:0) 1036 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15 1037 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0) 1038 Total transfers 0 (0 reads and 0 writes) 1039 1040 10411.6 Parallel port info in /proc/parport 1042--------------------------------------- 1043 1044The directory /proc/parport contains information about the parallel ports of 1045your system. It has one subdirectory for each port, named after the port 1046number (0,1,2,...). 1047 1048These directories contain the four files shown in Table 1-10. 1049 1050 1051Table 1-10: Files in /proc/parport 1052.............................................................................. 1053 File Content 1054 autoprobe Any IEEE-1284 device ID information that has been acquired. 1055 devices list of the device drivers using that port. A + will appear by the 1056 name of the device currently using the port (it might not appear 1057 against any). 1058 hardware Parallel port's base address, IRQ line and DMA channel. 1059 irq IRQ that parport is using for that port. This is in a separate 1060 file to allow you to alter it by writing a new value in (IRQ 1061 number or none). 1062.............................................................................. 1063 10641.7 TTY info in /proc/tty 1065------------------------- 1066 1067Information about the available and actually used tty's can be found in the 1068directory /proc/tty.You'll find entries for drivers and line disciplines in 1069this directory, as shown in Table 1-11. 1070 1071 1072Table 1-11: Files in /proc/tty 1073.............................................................................. 1074 File Content 1075 drivers list of drivers and their usage 1076 ldiscs registered line disciplines 1077 driver/serial usage statistic and status of single tty lines 1078.............................................................................. 1079 1080To see which tty's are currently in use, you can simply look into the file 1081/proc/tty/drivers: 1082 1083 > cat /proc/tty/drivers 1084 pty_slave /dev/pts 136 0-255 pty:slave 1085 pty_master /dev/ptm 128 0-255 pty:master 1086 pty_slave /dev/ttyp 3 0-255 pty:slave 1087 pty_master /dev/pty 2 0-255 pty:master 1088 serial /dev/cua 5 64-67 serial:callout 1089 serial /dev/ttyS 4 64-67 serial 1090 /dev/tty0 /dev/tty0 4 0 system:vtmaster 1091 /dev/ptmx /dev/ptmx 5 2 system 1092 /dev/console /dev/console 5 1 system:console 1093 /dev/tty /dev/tty 5 0 system:/dev/tty 1094 unknown /dev/tty 4 1-63 console 1095 1096 10971.8 Miscellaneous kernel statistics in /proc/stat 1098------------------------------------------------- 1099 1100Various pieces of information about kernel activity are available in the 1101/proc/stat file. All of the numbers reported in this file are aggregates 1102since the system first booted. For a quick look, simply cat the file: 1103 1104 > cat /proc/stat 1105 cpu 2255 34 2290 22625563 6290 127 456 0 0 1106 cpu0 1132 34 1441 11311718 3675 127 438 0 0 1107 cpu1 1123 0 849 11313845 2614 0 18 0 0 1108 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...] 1109 ctxt 1990473 1110 btime 1062191376 1111 processes 2915 1112 procs_running 1 1113 procs_blocked 0 1114 softirq 183433 0 21755 12 39 1137 231 21459 2263 1115 1116The very first "cpu" line aggregates the numbers in all of the other "cpuN" 1117lines. These numbers identify the amount of time the CPU has spent performing 1118different kinds of work. Time units are in USER_HZ (typically hundredths of a 1119second). The meanings of the columns are as follows, from left to right: 1120 1121- user: normal processes executing in user mode 1122- nice: niced processes executing in user mode 1123- system: processes executing in kernel mode 1124- idle: twiddling thumbs 1125- iowait: waiting for I/O to complete 1126- irq: servicing interrupts 1127- softirq: servicing softirqs 1128- steal: involuntary wait 1129- guest: running a normal guest 1130- guest_nice: running a niced guest 1131 1132The "intr" line gives counts of interrupts serviced since boot time, for each 1133of the possible system interrupts. The first column is the total of all 1134interrupts serviced; each subsequent column is the total for that particular 1135interrupt. 1136 1137The "ctxt" line gives the total number of context switches across all CPUs. 1138 1139The "btime" line gives the time at which the system booted, in seconds since 1140the Unix epoch. 1141 1142The "processes" line gives the number of processes and threads created, which 1143includes (but is not limited to) those created by calls to the fork() and 1144clone() system calls. 1145 1146The "procs_running" line gives the total number of threads that are 1147running or ready to run (i.e., the total number of runnable threads). 1148 1149The "procs_blocked" line gives the number of processes currently blocked, 1150waiting for I/O to complete. 1151 1152The "softirq" line gives counts of softirqs serviced since boot time, for each 1153of the possible system softirqs. The first column is the total of all 1154softirqs serviced; each subsequent column is the total for that particular 1155softirq. 1156 1157 11581.9 Ext4 file system parameters 1159------------------------------ 1160 1161Information about mounted ext4 file systems can be found in 1162/proc/fs/ext4. Each mounted filesystem will have a directory in 1163/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or 1164/proc/fs/ext4/dm-0). The files in each per-device directory are shown 1165in Table 1-12, below. 1166 1167Table 1-12: Files in /proc/fs/ext4/<devname> 1168.............................................................................. 1169 File Content 1170 mb_groups details of multiblock allocator buddy cache of free blocks 1171.............................................................................. 1172 1173 1174------------------------------------------------------------------------------ 1175Summary 1176------------------------------------------------------------------------------ 1177The /proc file system serves information about the running system. It not only 1178allows access to process data but also allows you to request the kernel status 1179by reading files in the hierarchy. 1180 1181The directory structure of /proc reflects the types of information and makes 1182it easy, if not obvious, where to look for specific data. 1183------------------------------------------------------------------------------ 1184 1185------------------------------------------------------------------------------ 1186CHAPTER 2: MODIFYING SYSTEM PARAMETERS 1187------------------------------------------------------------------------------ 1188 1189------------------------------------------------------------------------------ 1190In This Chapter 1191------------------------------------------------------------------------------ 1192* Modifying kernel parameters by writing into files found in /proc/sys 1193* Exploring the files which modify certain parameters 1194* Review of the /proc/sys file tree 1195------------------------------------------------------------------------------ 1196 1197 1198A very interesting part of /proc is the directory /proc/sys. This is not only 1199a source of information, it also allows you to change parameters within the 1200kernel. Be very careful when attempting this. You can optimize your system, 1201but you can also cause it to crash. Never alter kernel parameters on a 1202production system. Set up a development machine and test to make sure that 1203everything works the way you want it to. You may have no alternative but to 1204reboot the machine once an error has been made. 1205 1206To change a value, simply echo the new value into the file. An example is 1207given below in the section on the file system data. You need to be root to do 1208this. You can create your own boot script to perform this every time your 1209system boots. 1210 1211The files in /proc/sys can be used to fine tune and monitor miscellaneous and 1212general things in the operation of the Linux kernel. Since some of the files 1213can inadvertently disrupt your system, it is advisable to read both 1214documentation and source before actually making adjustments. In any case, be 1215very careful when writing to any of these files. The entries in /proc may 1216change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt 1217review the kernel documentation in the directory /usr/src/linux/Documentation. 1218This chapter is heavily based on the documentation included in the pre 2.2 1219kernels, and became part of it in version 2.2.1 of the Linux kernel. 1220 1221Please see: Documentation/sysctls/ directory for descriptions of these 1222entries. 1223 1224------------------------------------------------------------------------------ 1225Summary 1226------------------------------------------------------------------------------ 1227Certain aspects of kernel behavior can be modified at runtime, without the 1228need to recompile the kernel, or even to reboot the system. The files in the 1229/proc/sys tree can not only be read, but also modified. You can use the echo 1230command to write value into these files, thereby changing the default settings 1231of the kernel. 1232------------------------------------------------------------------------------ 1233 1234------------------------------------------------------------------------------ 1235CHAPTER 3: PER-PROCESS PARAMETERS 1236------------------------------------------------------------------------------ 1237 12383.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score 1239-------------------------------------------------------------------------------- 1240 1241These file can be used to adjust the badness heuristic used to select which 1242process gets killed in out of memory conditions. 1243 1244The badness heuristic assigns a value to each candidate task ranging from 0 1245(never kill) to 1000 (always kill) to determine which process is targeted. The 1246units are roughly a proportion along that range of allowed memory the process 1247may allocate from based on an estimation of its current memory and swap use. 1248For example, if a task is using all allowed memory, its badness score will be 12491000. If it is using half of its allowed memory, its score will be 500. 1250 1251There is an additional factor included in the badness score: root 1252processes are given 3% extra memory over other tasks. 1253 1254The amount of "allowed" memory depends on the context in which the oom killer 1255was called. If it is due to the memory assigned to the allocating task's cpuset 1256being exhausted, the allowed memory represents the set of mems assigned to that 1257cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed 1258memory represents the set of mempolicy nodes. If it is due to a memory 1259limit (or swap limit) being reached, the allowed memory is that configured 1260limit. Finally, if it is due to the entire system being out of memory, the 1261allowed memory represents all allocatable resources. 1262 1263The value of /proc/<pid>/oom_score_adj is added to the badness score before it 1264is used to determine which task to kill. Acceptable values range from -1000 1265(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to 1266polarize the preference for oom killing either by always preferring a certain 1267task or completely disabling it. The lowest possible value, -1000, is 1268equivalent to disabling oom killing entirely for that task since it will always 1269report a badness score of 0. 1270 1271Consequently, it is very simple for userspace to define the amount of memory to 1272consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for 1273example, is roughly equivalent to allowing the remainder of tasks sharing the 1274same system, cpuset, mempolicy, or memory controller resources to use at least 127550% more memory. A value of -500, on the other hand, would be roughly 1276equivalent to discounting 50% of the task's allowed memory from being considered 1277as scoring against the task. 1278 1279For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also 1280be used to tune the badness score. Its acceptable values range from -16 1281(OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17 1282(OOM_DISABLE) to disable oom killing entirely for that task. Its value is 1283scaled linearly with /proc/<pid>/oom_score_adj. 1284 1285Writing to /proc/<pid>/oom_score_adj or /proc/<pid>/oom_adj will change the 1286other with its scaled value. 1287 1288NOTICE: /proc/<pid>/oom_adj is deprecated and will be removed, please see 1289Documentation/feature-removal-schedule.txt. 1290 1291Caveat: when a parent task is selected, the oom killer will sacrifice any first 1292generation children with seperate address spaces instead, if possible. This 1293avoids servers and important system daemons from being killed and loses the 1294minimal amount of work. 1295 1296 12973.2 /proc/<pid>/oom_score - Display current oom-killer score 1298------------------------------------------------------------- 1299 1300This file can be used to check the current score used by the oom-killer is for 1301any given <pid>. Use it together with /proc/<pid>/oom_adj to tune which 1302process should be killed in an out-of-memory situation. 1303 1304 13053.3 /proc/<pid>/io - Display the IO accounting fields 1306------------------------------------------------------- 1307 1308This file contains IO statistics for each running process 1309 1310Example 1311------- 1312 1313test:/tmp # dd if=/dev/zero of=/tmp/test.dat & 1314[1] 3828 1315 1316test:/tmp # cat /proc/3828/io 1317rchar: 323934931 1318wchar: 323929600 1319syscr: 632687 1320syscw: 632675 1321read_bytes: 0 1322write_bytes: 323932160 1323cancelled_write_bytes: 0 1324 1325 1326Description 1327----------- 1328 1329rchar 1330----- 1331 1332I/O counter: chars read 1333The number of bytes which this task has caused to be read from storage. This 1334is simply the sum of bytes which this process passed to read() and pread(). 1335It includes things like tty IO and it is unaffected by whether or not actual 1336physical disk IO was required (the read might have been satisfied from 1337pagecache) 1338 1339 1340wchar 1341----- 1342 1343I/O counter: chars written 1344The number of bytes which this task has caused, or shall cause to be written 1345to disk. Similar caveats apply here as with rchar. 1346 1347 1348syscr 1349----- 1350 1351I/O counter: read syscalls 1352Attempt to count the number of read I/O operations, i.e. syscalls like read() 1353and pread(). 1354 1355 1356syscw 1357----- 1358 1359I/O counter: write syscalls 1360Attempt to count the number of write I/O operations, i.e. syscalls like 1361write() and pwrite(). 1362 1363 1364read_bytes 1365---------- 1366 1367I/O counter: bytes read 1368Attempt to count the number of bytes which this process really did cause to 1369be fetched from the storage layer. Done at the submit_bio() level, so it is 1370accurate for block-backed filesystems. <please add status regarding NFS and 1371CIFS at a later time> 1372 1373 1374write_bytes 1375----------- 1376 1377I/O counter: bytes written 1378Attempt to count the number of bytes which this process caused to be sent to 1379the storage layer. This is done at page-dirtying time. 1380 1381 1382cancelled_write_bytes 1383--------------------- 1384 1385The big inaccuracy here is truncate. If a process writes 1MB to a file and 1386then deletes the file, it will in fact perform no writeout. But it will have 1387been accounted as having caused 1MB of write. 1388In other words: The number of bytes which this process caused to not happen, 1389by truncating pagecache. A task can cause "negative" IO too. If this task 1390truncates some dirty pagecache, some IO which another task has been accounted 1391for (in its write_bytes) will not be happening. We _could_ just subtract that 1392from the truncating task's write_bytes, but there is information loss in doing 1393that. 1394 1395 1396Note 1397---- 1398 1399At its current implementation state, this is a bit racy on 32-bit machines: if 1400process A reads process B's /proc/pid/io while process B is updating one of 1401those 64-bit counters, process A could see an intermediate result. 1402 1403 1404More information about this can be found within the taskstats documentation in 1405Documentation/accounting. 1406 14073.4 /proc/<pid>/coredump_filter - Core dump filtering settings 1408--------------------------------------------------------------- 1409When a process is dumped, all anonymous memory is written to a core file as 1410long as the size of the core file isn't limited. But sometimes we don't want 1411to dump some memory segments, for example, huge shared memory. Conversely, 1412sometimes we want to save file-backed memory segments into a core file, not 1413only the individual files. 1414 1415/proc/<pid>/coredump_filter allows you to customize which memory segments 1416will be dumped when the <pid> process is dumped. coredump_filter is a bitmask 1417of memory types. If a bit of the bitmask is set, memory segments of the 1418corresponding memory type are dumped, otherwise they are not dumped. 1419 1420The following 7 memory types are supported: 1421 - (bit 0) anonymous private memory 1422 - (bit 1) anonymous shared memory 1423 - (bit 2) file-backed private memory 1424 - (bit 3) file-backed shared memory 1425 - (bit 4) ELF header pages in file-backed private memory areas (it is 1426 effective only if the bit 2 is cleared) 1427 - (bit 5) hugetlb private memory 1428 - (bit 6) hugetlb shared memory 1429 1430 Note that MMIO pages such as frame buffer are never dumped and vDSO pages 1431 are always dumped regardless of the bitmask status. 1432 1433 Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only 1434 effected by bit 5-6. 1435 1436Default value of coredump_filter is 0x23; this means all anonymous memory 1437segments and hugetlb private memory are dumped. 1438 1439If you don't want to dump all shared memory segments attached to pid 1234, 1440write 0x21 to the process's proc file. 1441 1442 $ echo 0x21 > /proc/1234/coredump_filter 1443 1444When a new process is created, the process inherits the bitmask status from its 1445parent. It is useful to set up coredump_filter before the program runs. 1446For example: 1447 1448 $ echo 0x7 > /proc/self/coredump_filter 1449 $ ./some_program 1450 14513.5 /proc/<pid>/mountinfo - Information about mounts 1452-------------------------------------------------------- 1453 1454This file contains lines of the form: 1455 145636 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue 1457(1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11) 1458 1459(1) mount ID: unique identifier of the mount (may be reused after umount) 1460(2) parent ID: ID of parent (or of self for the top of the mount tree) 1461(3) major:minor: value of st_dev for files on filesystem 1462(4) root: root of the mount within the filesystem 1463(5) mount point: mount point relative to the process's root 1464(6) mount options: per mount options 1465(7) optional fields: zero or more fields of the form "tag[:value]" 1466(8) separator: marks the end of the optional fields 1467(9) filesystem type: name of filesystem of the form "type[.subtype]" 1468(10) mount source: filesystem specific information or "none" 1469(11) super options: per super block options 1470 1471Parsers should ignore all unrecognised optional fields. Currently the 1472possible optional fields are: 1473 1474shared:X mount is shared in peer group X 1475master:X mount is slave to peer group X 1476propagate_from:X mount is slave and receives propagation from peer group X (*) 1477unbindable mount is unbindable 1478 1479(*) X is the closest dominant peer group under the process's root. If 1480X is the immediate master of the mount, or if there's no dominant peer 1481group under the same root, then only the "master:X" field is present 1482and not the "propagate_from:X" field. 1483 1484For more information on mount propagation see: 1485 1486 Documentation/filesystems/sharedsubtree.txt 1487 1488 14893.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 1490-------------------------------------------------------- 1491These files provide a method to access a tasks comm value. It also allows for 1492a task to set its own or one of its thread siblings comm value. The comm value 1493is limited in size compared to the cmdline value, so writing anything longer 1494then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated 1495comm value. 1496