1.. SPDX-License-Identifier: GPL-2.0 2 3========================== 4General Filesystem Caching 5========================== 6 7Overview 8======== 9 10This facility is a general purpose cache for network filesystems, though it 11could be used for caching other things such as ISO9660 filesystems too. 12 13FS-Cache mediates between cache backends (such as CacheFS) and network 14filesystems:: 15 16 +---------+ 17 | | +--------------+ 18 | NFS |--+ | | 19 | | | +-->| CacheFS | 20 +---------+ | +----------+ | | /dev/hda5 | 21 | | | | +--------------+ 22 +---------+ +-->| | | 23 | | | |--+ 24 | AFS |----->| FS-Cache | 25 | | | |--+ 26 +---------+ +-->| | | 27 | | | | +--------------+ 28 +---------+ | +----------+ | | | 29 | | | +-->| CacheFiles | 30 | ISOFS |--+ | /var/cache | 31 | | +--------------+ 32 +---------+ 33 34Or to look at it another way, FS-Cache is a module that provides a caching 35facility to a network filesystem such that the cache is transparent to the 36user:: 37 38 +---------+ 39 | | 40 | Server | 41 | | 42 +---------+ 43 | NETWORK 44 ~~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 45 | 46 | +----------+ 47 V | | 48 +---------+ | | 49 | | | | 50 | NFS |----->| FS-Cache | 51 | | | |--+ 52 +---------+ | | | +--------------+ +--------------+ 53 | | | | | | | | 54 V +----------+ +-->| CacheFiles |-->| Ext3 | 55 +---------+ | /var/cache | | /dev/sda6 | 56 | | +--------------+ +--------------+ 57 | VFS | ^ ^ 58 | | | | 59 +---------+ +--------------+ | 60 | KERNEL SPACE | | 61 ~~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~|~~~~~~|~~~~ 62 | USER SPACE | | 63 V | | 64 +---------+ +--------------+ 65 | | | | 66 | Process | | cachefilesd | 67 | | | | 68 +---------+ +--------------+ 69 70 71FS-Cache does not follow the idea of completely loading every netfs file 72opened in its entirety into a cache before permitting it to be accessed and 73then serving the pages out of that cache rather than the netfs inode because: 74 75 (1) It must be practical to operate without a cache. 76 77 (2) The size of any accessible file must not be limited to the size of the 78 cache. 79 80 (3) The combined size of all opened files (this includes mapped libraries) 81 must not be limited to the size of the cache. 82 83 (4) The user should not be forced to download an entire file just to do a 84 one-off access of a small portion of it (such as might be done with the 85 "file" program). 86 87It instead serves the cache out in PAGE_SIZE chunks as and when requested by 88the netfs('s) using it. 89 90 91FS-Cache provides the following facilities: 92 93 (1) More than one cache can be used at once. Caches can be selected 94 explicitly by use of tags. 95 96 (2) Caches can be added / removed at any time. 97 98 (3) The netfs is provided with an interface that allows either party to 99 withdraw caching facilities from a file (required for (2)). 100 101 (4) The interface to the netfs returns as few errors as possible, preferring 102 rather to let the netfs remain oblivious. 103 104 (5) Cookies are used to represent indices, files and other objects to the 105 netfs. The simplest cookie is just a NULL pointer - indicating nothing 106 cached there. 107 108 (6) The netfs is allowed to propose - dynamically - any index hierarchy it 109 desires, though it must be aware that the index search function is 110 recursive, stack space is limited, and indices can only be children of 111 indices. 112 113 (7) Data I/O is done direct to and from the netfs's pages. The netfs 114 indicates that page A is at index B of the data-file represented by cookie 115 C, and that it should be read or written. The cache backend may or may 116 not start I/O on that page, but if it does, a netfs callback will be 117 invoked to indicate completion. The I/O may be either synchronous or 118 asynchronous. 119 120 (8) Cookies can be "retired" upon release. At this point FS-Cache will mark 121 them as obsolete and the index hierarchy rooted at that point will get 122 recycled. 123 124 (9) The netfs provides a "match" function for index searches. In addition to 125 saying whether a match was made or not, this can also specify that an 126 entry should be updated or deleted. 127 128(10) As much as possible is done asynchronously. 129 130 131FS-Cache maintains a virtual indexing tree in which all indices, files, objects 132and pages are kept. Bits of this tree may actually reside in one or more 133caches:: 134 135 FSDEF 136 | 137 +------------------------------------+ 138 | | 139 NFS AFS 140 | | 141 +--------------------------+ +-----------+ 142 | | | | 143 homedir mirror afs.org redhat.com 144 | | | 145 +------------+ +---------------+ +----------+ 146 | | | | | | 147 00001 00002 00007 00125 vol00001 vol00002 148 | | | | | 149 +---+---+ +-----+ +---+ +------+------+ +-----+----+ 150 | | | | | | | | | | | | | 151 PG0 PG1 PG2 PG0 XATTR PG0 PG1 DIRENT DIRENT DIRENT R/W R/O Bak 152 | | 153 PG0 +-------+ 154 | | 155 00001 00003 156 | 157 +---+---+ 158 | | | 159 PG0 PG1 PG2 160 161In the example above, you can see two netfs's being backed: NFS and AFS. These 162have different index hierarchies: 163 164 * The NFS primary index contains per-server indices. Each server index is 165 indexed by NFS file handles to get data file objects. Each data file 166 objects can have an array of pages, but may also have further child 167 objects, such as extended attributes and directory entries. Extended 168 attribute objects themselves have page-array contents. 169 170 * The AFS primary index contains per-cell indices. Each cell index contains 171 per-logical-volume indices. Each of volume index contains up to three 172 indices for the read-write, read-only and backup mirrors of those volumes. 173 Each of these contains vnode data file objects, each of which contains an 174 array of pages. 175 176The very top index is the FS-Cache master index in which individual netfs's 177have entries. 178 179Any index object may reside in more than one cache, provided it only has index 180children. Any index with non-index object children will be assumed to only 181reside in one cache. 182 183 184The netfs API to FS-Cache can be found in: 185 186 Documentation/filesystems/caching/netfs-api.rst 187 188The cache backend API to FS-Cache can be found in: 189 190 Documentation/filesystems/caching/backend-api.rst 191 192A description of the internal representations and object state machine can be 193found in: 194 195 Documentation/filesystems/caching/object.rst 196 197 198Statistical Information 199======================= 200 201If FS-Cache is compiled with the following options enabled:: 202 203 CONFIG_FSCACHE_STATS=y 204 CONFIG_FSCACHE_HISTOGRAM=y 205 206then it will gather certain statistics and display them through a number of 207proc files. 208 209/proc/fs/fscache/stats 210---------------------- 211 212 This shows counts of a number of events that can happen in FS-Cache: 213 214+--------------+-------+-------------------------------------------------------+ 215|CLASS |EVENT |MEANING | 216+==============+=======+=======================================================+ 217|Cookies |idx=N |Number of index cookies allocated | 218+ +-------+-------------------------------------------------------+ 219| |dat=N |Number of data storage cookies allocated | 220+ +-------+-------------------------------------------------------+ 221| |spc=N |Number of special cookies allocated | 222+--------------+-------+-------------------------------------------------------+ 223|Objects |alc=N |Number of objects allocated | 224+ +-------+-------------------------------------------------------+ 225| |nal=N |Number of object allocation failures | 226+ +-------+-------------------------------------------------------+ 227| |avl=N |Number of objects that reached the available state | 228+ +-------+-------------------------------------------------------+ 229| |ded=N |Number of objects that reached the dead state | 230+--------------+-------+-------------------------------------------------------+ 231|ChkAux |non=N |Number of objects that didn't have a coherency check | 232+ +-------+-------------------------------------------------------+ 233| |ok=N |Number of objects that passed a coherency check | 234+ +-------+-------------------------------------------------------+ 235| |upd=N |Number of objects that needed a coherency data update | 236+ +-------+-------------------------------------------------------+ 237| |obs=N |Number of objects that were declared obsolete | 238+--------------+-------+-------------------------------------------------------+ 239|Pages |mrk=N |Number of pages marked as being cached | 240| |unc=N |Number of uncache page requests seen | 241+--------------+-------+-------------------------------------------------------+ 242|Acquire |n=N |Number of acquire cookie requests seen | 243+ +-------+-------------------------------------------------------+ 244| |nul=N |Number of acq reqs given a NULL parent | 245+ +-------+-------------------------------------------------------+ 246| |noc=N |Number of acq reqs rejected due to no cache available | 247+ +-------+-------------------------------------------------------+ 248| |ok=N |Number of acq reqs succeeded | 249+ +-------+-------------------------------------------------------+ 250| |nbf=N |Number of acq reqs rejected due to error | 251+ +-------+-------------------------------------------------------+ 252| |oom=N |Number of acq reqs failed on ENOMEM | 253+--------------+-------+-------------------------------------------------------+ 254|Lookups |n=N |Number of lookup calls made on cache backends | 255+ +-------+-------------------------------------------------------+ 256| |neg=N |Number of negative lookups made | 257+ +-------+-------------------------------------------------------+ 258| |pos=N |Number of positive lookups made | 259+ +-------+-------------------------------------------------------+ 260| |crt=N |Number of objects created by lookup | 261+ +-------+-------------------------------------------------------+ 262| |tmo=N |Number of lookups timed out and requeued | 263+--------------+-------+-------------------------------------------------------+ 264|Updates |n=N |Number of update cookie requests seen | 265+ +-------+-------------------------------------------------------+ 266| |nul=N |Number of upd reqs given a NULL parent | 267+ +-------+-------------------------------------------------------+ 268| |run=N |Number of upd reqs granted CPU time | 269+--------------+-------+-------------------------------------------------------+ 270|Relinqs |n=N |Number of relinquish cookie requests seen | 271+ +-------+-------------------------------------------------------+ 272| |nul=N |Number of rlq reqs given a NULL parent | 273+ +-------+-------------------------------------------------------+ 274| |wcr=N |Number of rlq reqs waited on completion of creation | 275+--------------+-------+-------------------------------------------------------+ 276|AttrChg |n=N |Number of attribute changed requests seen | 277+ +-------+-------------------------------------------------------+ 278| |ok=N |Number of attr changed requests queued | 279+ +-------+-------------------------------------------------------+ 280| |nbf=N |Number of attr changed rejected -ENOBUFS | 281+ +-------+-------------------------------------------------------+ 282| |oom=N |Number of attr changed failed -ENOMEM | 283+ +-------+-------------------------------------------------------+ 284| |run=N |Number of attr changed ops given CPU time | 285+--------------+-------+-------------------------------------------------------+ 286|Allocs |n=N |Number of allocation requests seen | 287+ +-------+-------------------------------------------------------+ 288| |ok=N |Number of successful alloc reqs | 289+ +-------+-------------------------------------------------------+ 290| |wt=N |Number of alloc reqs that waited on lookup completion | 291+ +-------+-------------------------------------------------------+ 292| |nbf=N |Number of alloc reqs rejected -ENOBUFS | 293+ +-------+-------------------------------------------------------+ 294| |int=N |Number of alloc reqs aborted -ERESTARTSYS | 295+ +-------+-------------------------------------------------------+ 296| |ops=N |Number of alloc reqs submitted | 297+ +-------+-------------------------------------------------------+ 298| |owt=N |Number of alloc reqs waited for CPU time | 299+ +-------+-------------------------------------------------------+ 300| |abt=N |Number of alloc reqs aborted due to object death | 301+--------------+-------+-------------------------------------------------------+ 302|Retrvls |n=N |Number of retrieval (read) requests seen | 303+ +-------+-------------------------------------------------------+ 304| |ok=N |Number of successful retr reqs | 305+ +-------+-------------------------------------------------------+ 306| |wt=N |Number of retr reqs that waited on lookup completion | 307+ +-------+-------------------------------------------------------+ 308| |nod=N |Number of retr reqs returned -ENODATA | 309+ +-------+-------------------------------------------------------+ 310| |nbf=N |Number of retr reqs rejected -ENOBUFS | 311+ +-------+-------------------------------------------------------+ 312| |int=N |Number of retr reqs aborted -ERESTARTSYS | 313+ +-------+-------------------------------------------------------+ 314| |oom=N |Number of retr reqs failed -ENOMEM | 315+ +-------+-------------------------------------------------------+ 316| |ops=N |Number of retr reqs submitted | 317+ +-------+-------------------------------------------------------+ 318| |owt=N |Number of retr reqs waited for CPU time | 319+ +-------+-------------------------------------------------------+ 320| |abt=N |Number of retr reqs aborted due to object death | 321+--------------+-------+-------------------------------------------------------+ 322|Stores |n=N |Number of storage (write) requests seen | 323+ +-------+-------------------------------------------------------+ 324| |ok=N |Number of successful store reqs | 325+ +-------+-------------------------------------------------------+ 326| |agn=N |Number of store reqs on a page already pending storage | 327+ +-------+-------------------------------------------------------+ 328| |nbf=N |Number of store reqs rejected -ENOBUFS | 329+ +-------+-------------------------------------------------------+ 330| |oom=N |Number of store reqs failed -ENOMEM | 331+ +-------+-------------------------------------------------------+ 332| |ops=N |Number of store reqs submitted | 333+ +-------+-------------------------------------------------------+ 334| |run=N |Number of store reqs granted CPU time | 335+ +-------+-------------------------------------------------------+ 336| |pgs=N |Number of pages given store req processing time | 337+ +-------+-------------------------------------------------------+ 338| |rxd=N |Number of store reqs deleted from tracking tree | 339+ +-------+-------------------------------------------------------+ 340| |olm=N |Number of store reqs over store limit | 341+--------------+-------+-------------------------------------------------------+ 342|VmScan |nos=N |Number of release reqs against pages with no | 343| | |pending store | 344+ +-------+-------------------------------------------------------+ 345| |gon=N |Number of release reqs against pages stored by | 346| | |time lock granted | 347+ +-------+-------------------------------------------------------+ 348| |bsy=N |Number of release reqs ignored due to in-progress store| 349+ +-------+-------------------------------------------------------+ 350| |can=N |Number of page stores cancelled due to release req | 351+--------------+-------+-------------------------------------------------------+ 352|Ops |pend=N |Number of times async ops added to pending queues | 353+ +-------+-------------------------------------------------------+ 354| |run=N |Number of times async ops given CPU time | 355+ +-------+-------------------------------------------------------+ 356| |enq=N |Number of times async ops queued for processing | 357+ +-------+-------------------------------------------------------+ 358| |can=N |Number of async ops cancelled | 359+ +-------+-------------------------------------------------------+ 360| |rej=N |Number of async ops rejected due to object | 361| | |lookup/create failure | 362+ +-------+-------------------------------------------------------+ 363| |ini=N |Number of async ops initialised | 364+ +-------+-------------------------------------------------------+ 365| |dfr=N |Number of async ops queued for deferred release | 366+ +-------+-------------------------------------------------------+ 367| |rel=N |Number of async ops released | 368| | |(should equal ini=N when idle) | 369+ +-------+-------------------------------------------------------+ 370| |gc=N |Number of deferred-release async ops garbage collected | 371+--------------+-------+-------------------------------------------------------+ 372|CacheOp |alo=N |Number of in-progress alloc_object() cache ops | 373+ +-------+-------------------------------------------------------+ 374| |luo=N |Number of in-progress lookup_object() cache ops | 375+ +-------+-------------------------------------------------------+ 376| |luc=N |Number of in-progress lookup_complete() cache ops | 377+ +-------+-------------------------------------------------------+ 378| |gro=N |Number of in-progress grab_object() cache ops | 379+ +-------+-------------------------------------------------------+ 380| |upo=N |Number of in-progress update_object() cache ops | 381+ +-------+-------------------------------------------------------+ 382| |dro=N |Number of in-progress drop_object() cache ops | 383+ +-------+-------------------------------------------------------+ 384| |pto=N |Number of in-progress put_object() cache ops | 385+ +-------+-------------------------------------------------------+ 386| |syn=N |Number of in-progress sync_cache() cache ops | 387+ +-------+-------------------------------------------------------+ 388| |atc=N |Number of in-progress attr_changed() cache ops | 389+ +-------+-------------------------------------------------------+ 390| |rap=N |Number of in-progress read_or_alloc_page() cache ops | 391+ +-------+-------------------------------------------------------+ 392| |ras=N |Number of in-progress read_or_alloc_pages() cache ops | 393+ +-------+-------------------------------------------------------+ 394| |alp=N |Number of in-progress allocate_page() cache ops | 395+ +-------+-------------------------------------------------------+ 396| |als=N |Number of in-progress allocate_pages() cache ops | 397+ +-------+-------------------------------------------------------+ 398| |wrp=N |Number of in-progress write_page() cache ops | 399+ +-------+-------------------------------------------------------+ 400| |ucp=N |Number of in-progress uncache_page() cache ops | 401+ +-------+-------------------------------------------------------+ 402| |dsp=N |Number of in-progress dissociate_pages() cache ops | 403+--------------+-------+-------------------------------------------------------+ 404|CacheEv |nsp=N |Number of object lookups/creations rejected due to | 405| | |lack of space | 406+ +-------+-------------------------------------------------------+ 407| |stl=N |Number of stale objects deleted | 408+ +-------+-------------------------------------------------------+ 409| |rtr=N |Number of objects retired when relinquished | 410+ +-------+-------------------------------------------------------+ 411| |cul=N |Number of objects culled | 412+--------------+-------+-------------------------------------------------------+ 413 414 415 416/proc/fs/fscache/histogram 417-------------------------- 418 419 :: 420 421 cat /proc/fs/fscache/histogram 422 JIFS SECS OBJ INST OP RUNS OBJ RUNS RETRV DLY RETRIEVLS 423 ===== ===== ========= ========= ========= ========= ========= 424 425 This shows the breakdown of the number of times each amount of time 426 between 0 jiffies and HZ-1 jiffies a variety of tasks took to run. The 427 columns are as follows: 428 429 ========= ======================================================= 430 COLUMN TIME MEASUREMENT 431 ========= ======================================================= 432 OBJ INST Length of time to instantiate an object 433 OP RUNS Length of time a call to process an operation took 434 OBJ RUNS Length of time a call to process an object event took 435 RETRV DLY Time between an requesting a read and lookup completing 436 RETRIEVLS Time between beginning and end of a retrieval 437 ========= ======================================================= 438 439 Each row shows the number of events that took a particular range of times. 440 Each step is 1 jiffy in size. The JIFS column indicates the particular 441 jiffy range covered, and the SECS field the equivalent number of seconds. 442 443 444 445Object List 446=========== 447 448If CONFIG_FSCACHE_OBJECT_LIST is enabled, the FS-Cache facility will maintain a 449list of all the objects currently allocated and allow them to be viewed 450through:: 451 452 /proc/fs/fscache/objects 453 454This will look something like:: 455 456 [root@andromeda ~]# head /proc/fs/fscache/objects 457 OBJECT PARENT STAT CHLDN OPS OOP IPR EX READS EM EV F S | NETFS_COOKIE_DEF TY FL NETFS_DATA OBJECT_KEY, AUX_DATA 458 ======== ======== ==== ===== === === === == ===== == == = = | ================ == == ================ ================ 459 17e4b 2 ACTV 0 0 0 0 0 0 7b 4 0 0 | NFS.fh DT 0 ffff88001dd82820 010006017edcf8bbc93b43298fdfbe71e50b57b13a172c0117f38472, e567634700000000000000000000000063f2404a000000000000000000000000c9030000000000000000000063f2404a 460 1693a 2 ACTV 0 0 0 0 0 0 7b 4 0 0 | NFS.fh DT 0 ffff88002db23380 010006017edcf8bbc93b43298fdfbe71e50b57b1e0162c01a2df0ea6, 420ebc4a000000000000000000000000420ebc4a0000000000000000000000000e1801000000000000000000420ebc4a 461 462where the first set of columns before the '|' describe the object: 463 464 ======= =============================================================== 465 COLUMN DESCRIPTION 466 ======= =============================================================== 467 OBJECT Object debugging ID (appears as OBJ%x in some debug messages) 468 PARENT Debugging ID of parent object 469 STAT Object state 470 CHLDN Number of child objects of this object 471 OPS Number of outstanding operations on this object 472 OOP Number of outstanding child object management operations 473 IPR 474 EX Number of outstanding exclusive operations 475 READS Number of outstanding read operations 476 EM Object's event mask 477 EV Events raised on this object 478 F Object flags 479 S Object work item busy state mask (1:pending 2:running) 480 ======= =============================================================== 481 482and the second set of columns describe the object's cookie, if present: 483 484 ================ ====================================================== 485 COLUMN DESCRIPTION 486 ================ ====================================================== 487 NETFS_COOKIE_DEF Name of netfs cookie definition 488 TY Cookie type (IX - index, DT - data, hex - special) 489 FL Cookie flags 490 NETFS_DATA Netfs private data stored in the cookie 491 OBJECT_KEY Object key } 1 column, with separating comma 492 AUX_DATA Object aux data } presence may be configured 493 ================ ====================================================== 494 495The data shown may be filtered by attaching the a key to an appropriate keyring 496before viewing the file. Something like:: 497 498 keyctl add user fscache:objlist <restrictions> @s 499 500where <restrictions> are a selection of the following letters: 501 502 == ========================================================= 503 K Show hexdump of object key (don't show if not given) 504 A Show hexdump of object aux data (don't show if not given) 505 == ========================================================= 506 507and the following paired letters: 508 509 == ========================================================= 510 C Show objects that have a cookie 511 c Show objects that don't have a cookie 512 B Show objects that are busy 513 b Show objects that aren't busy 514 W Show objects that have pending writes 515 w Show objects that don't have pending writes 516 R Show objects that have outstanding reads 517 r Show objects that don't have outstanding reads 518 S Show objects that have work queued 519 s Show objects that don't have work queued 520 == ========================================================= 521 522If neither side of a letter pair is given, then both are implied. For example: 523 524 keyctl add user fscache:objlist KB @s 525 526shows objects that are busy, and lists their object keys, but does not dump 527their auxiliary data. It also implies "CcWwRrSs", but as 'B' is given, 'b' is 528not implied. 529 530By default all objects and all fields will be shown. 531 532 533Debugging 534========= 535 536If CONFIG_FSCACHE_DEBUG is enabled, the FS-Cache facility can have runtime 537debugging enabled by adjusting the value in:: 538 539 /sys/module/fscache/parameters/debug 540 541This is a bitmask of debugging streams to enable: 542 543 ======= ======= =============================== ======================= 544 BIT VALUE STREAM POINT 545 ======= ======= =============================== ======================= 546 0 1 Cache management Function entry trace 547 1 2 Function exit trace 548 2 4 General 549 3 8 Cookie management Function entry trace 550 4 16 Function exit trace 551 5 32 General 552 6 64 Page handling Function entry trace 553 7 128 Function exit trace 554 8 256 General 555 9 512 Operation management Function entry trace 556 10 1024 Function exit trace 557 11 2048 General 558 ======= ======= =============================== ======================= 559 560The appropriate set of values should be OR'd together and the result written to 561the control file. For example:: 562 563 echo $((1|8|64)) >/sys/module/fscache/parameters/debug 564 565will turn on all function entry debugging. 566